End of month, anniversary, or something else entirely? Due dates for national renewal fees across the EPC states

Posted on August 2, 2026 by Dr. Dirk Franke

As long as the European application is pending before the EPO, a single clear rule applies. On grant, that clarity fragments into 39 national regimes — and they diverge on precisely the point most easily overlooked in docketing: the due date itself.

The starting point: Rule 51(1) EPC as a deceptive habit

For the European patent application the position is unambiguous. Under Rule 51(1) EPC, a renewal fee falls due “on the last day of the month containing the anniversary of the date of filing of the European patent application”. A filing date of 9 October means a due date of 31 October. Where the filing date falls on 31 May, the due date remains 31 May. The renewal fee for the third year is the first payable (Art. 86(1) EPC) and falls due on the last day of the month containing the second anniversary of filing.

The rule is generous: on average it hands the applicant a little over two weeks. And that is exactly where the trap lies. Anyone who has worked to the end of the month for years tends to carry that habit unexamined into the national phases after grant — and is wrong in roughly a third of the contracting states.

The break at grant: Art. 141 EPC

The obligation to pay renewal fees to the EPO terminates with the payment of the fee due for the year in which the mention of grant is published (Art. 86(2) EPC). Under Art. 141 EPC, national renewal fees may only be imposed for the years following that year. From that moment national law governs exclusively — including the national due date rule, the national period of grace and the national surcharge.

The European bundle patent is therefore also a bundle of deadlines. Validate in ten states and, from grant onwards, you are administering ten deadlines which all derive from the same filing date but do not all fall on the same day.

Group 1: due on the last day of the month of filing

The majority of contracting states follow the pattern of Rule 51(1) EPC. On the sources reviewed, these are:

Albania, Austria, Belgium, Denmark, Estonia, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Latvia, Liechtenstein, Luxembourg, Malta, Monaco, the Netherlands, Norway, San Marino, Spain, Sweden, Switzerland and the United Kingdom.

For Germany this follows directly from § 3(2) PatKostG: the renewal fee falls due “on expiry of the month whose designation corresponds to the month into which the filing date falls”. Austria applies the same approach, the Austrian Patent Office expressly referring to the last day of the month of filing. For France the same result follows from Art. R613-46 CPI; for Switzerland and Liechtenstein from the common territory of protection under the Patent Protection Treaty.

What is striking about this group is how heterogeneous it is. It contains both the classic high-volume validation states and micro-states such as Monaco and San Marino. There is no geographic or legal-family pattern to fall back on.

Group 2: due on the anniversary of the filing date

The second group works to the calendar anniversary — so the deadline expires, on average, a good two weeks earlier than in Group 1. On the sources reviewed, these are:

Bulgaria, Croatia, the Czech Republic, Hungary, Montenegro, North Macedonia, Poland, Portugal, Romania, Serbia, Slovakia, Slovenia and Türkiye.

For Montenegro this follows directly from the statute: Art. 81(2) of the Montenegrin Law on Patents (Official Gazette No. 42/2015, 2/2017, 146/2021, 3/2023) provides that the fees are payable for the third year and each subsequent year and “shall be due on the anniversary of the date of filing of the application”. Art. 81(4) grants an additional period of six months against payment of a surcharge.

The concentration in Central, Eastern and South-Eastern Europe is notable. Portugal and Türkiye are the outliers, to the west and south-east respectively. A portfolio validated predominantly in Western Europe encounters this group rarely — and is caught out by it all the more reliably as soon as a Polish or Czech validation is added.

Two cases that fit neither group

Cyprus. Cyprus has no due date in the continental sense at all. Rule 41(1)(a) of the Patent Regulations of 1998 does not fix a date on which the fee becomes due; it fixes a window within which it must be paid: Form P.13 together with the annual fee is to be filed “within three months before the expiration” of the current year of protection, counted from the filing date. For validated European patents this applies by analogy under Rule 55(1), the years being counted from the filing date of the European application (Rule 55(2)).

Two consequences follow. First, the payment period ends with the expiry of the current year of protection — and that is the day before the anniversary. A right filed on 1 February 2020 completed its first year of protection on 31 January 2021; the widely quoted “due the day before the anniversary” is therefore correct as a docketing date, but it describes the legal effect rather than the drafting technique. Second, there is a three-month payment window: payment cannot validly be made more than three months before the year of protection expires. Paying unusually far in advance risks not lateness but prematurity. The period of grace is six months against a surcharge (Rule 41(1)(b)).

Lithuania. Lithuania follows a different scheme altogether: the maintenance fee is payable within the final two months of the current year of protection, rather than on a date derived from the filing date. Secondary sources are inconsistent here; the Lithuanian Patent Law should be checked before the deadline is docketed.

A point of terminology

The commonly heard formulation that in certain states the renewal fee is payable “on the filing date” is misleading and should be kept out of docketing instructions. What is meant is always the anniversary of the filing date — the same calendar day in a later year. No EPC contracting state imposes a payment obligation on the filing date itself; the first renewal fee falls due at the earliest in the second, and usually in the third to fifth year of protection. The risk of confusion is not merely linguistic: writing “filing date” instead of “anniversary of the filing date” into a docketing instruction produces miscalculations as soon as the entry is processed automatically.

Overview: renewal fee due dates in the 39 EPC contracting states

CodeStateRenewal fee due date
ALAlbaniaLast day of the month of filing
ATAustriaLast day of the month of filing
BEBelgiumLast day of the month of filing
BGBulgariaAnniversary of the filing date
CHSwitzerlandLast day of the month of filing
CYCyprusBefore expiry of the current year of protection = day before the anniversary; three-month payment window (Rule 41(1)(a) Patent Regs 1998)
CZCzech RepublicAnniversary of the filing date
DEGermanyLast day of the month of filing (§ 3(2) PatKostG)
DKDenmarkLast day of the month of filing
EEEstoniaLast day of the month of filing
ESSpainLast day of the month of filing
FIFinlandLast day of the month of filing
FRFranceLast day of the month of filing (Art. R613-46 CPI)
GBUnited KingdomLast day of the month of filing
GRGreeceLast day of the month of filing
HRCroatiaAnniversary of the filing date
HUHungaryAnniversary of the filing date
IEIrelandLast day of the month of filing
ISIcelandLast day of the month of filing
ITItalyLast day of the month of filing
LILiechtensteinLast day of the month of filing (common territory with CH)
LTLithuaniaDifferent scheme: final two months of the current year of protection — check separately
LULuxembourgLast day of the month of filing
LVLatviaLast day of the month of filing
MCMonacoLast day of the month of filing
MEMontenegroAnniversary of the filing date (Art. 81(2) Law on Patents)
MKNorth MacedoniaAnniversary of the filing date
MTMaltaLast day of the month of filing
NLNetherlandsLast day of the month of filing
NONorwayLast day of the month of filing
PLPolandAnniversary of the filing date
PTPortugalAnniversary of the filing date
RORomaniaAnniversary of the filing date
RSSerbiaAnniversary of the filing date
SESwedenLast day of the month of filing
SISloveniaAnniversary of the filing date
SKSlovakiaAnniversary of the filing date
SMSan MarinoLast day of the month of filing
TRTürkiyeAnniversary of the filing date

The practical consequence: work to the earlier date

One simple and robust working rule follows from this split — and it is the rule the CMS patent formalities guide for Europe recommends as well: after grant, docket to the anniversary of the filing date throughout, not to the end of the month. In the Group 1 states this forgoes an average of a little over two weeks — a price that does not begin to weigh against the risk of missing a due date in Group 2.

The single exception to this rule of thumb is Cyprus, where the preceding day already governs. Anyone with Cyprus in the portfolio must docket that deadline separately.

The special case: grant shortly after the anniversary

A distinct source of error arises where the mention of grant is published shortly after the anniversary of the filing date. In that constellation the first national renewal fee falls due in many states two months from the date of grant — potentially before the three-month validation period expires. Some states allow additional time for payment; by no means all do. In practice this means: if grant falls into that window, country selection for validation should be brought forward so that the first renewal fee does not slip into a surcharge-bearing grace period.

The second question: where must a national representative be appointed?

This question is often asked in the same breath as the due date question, but it operates on a different level and needs to be separated cleanly into three stages.

Stage 1 — proceedings before the EPO

Under Art. 133(2) EPC, natural and legal persons having neither their residence nor their principal place of business in an EPC contracting state must be represented by a professional representative (Art. 134 EPC) in all proceedings before the EPO, with the single exception of filing the European patent application. Legal persons having their seat in a contracting state may act through an employee under Art. 133(3) EPC. For parties established within the EPC territory there is therefore no compulsory representation before the EPO.

Stage 2 — validation and national proceedings

On grant, national procedural law takes over. The relevant connecting factor shifts: what matters is no longer establishment within the EPC territory, but as a rule establishment in the particular state or within the EEA.

For Germany this is governed by § 25 PatG: a person having neither residence, seat nor establishment in Germany may take part in proceedings before the DPMA and the Federal Patent Court, and may assert rights under the patent, only if they have appointed a lawyer (Rechtsanwalt) or patent attorney (Patentanwalt) as domestic representative. That representative is at the same time authorised to act in civil litigation concerning the patent. Parties established in Germany are under no such obligation.

A de facto obligation to instruct a local representative additionally exists wherever validation itself can only be effected through an admitted national representative. This applies in particular to the states requiring a full translation of the specification: Austria (unless granted in German), Bulgaria, Cyprus, the Czech Republic, Estonia, Greece, Italy, Poland, Portugal, Romania, San Marino, Serbia, Slovakia, Spain and Türkiye. In these states instructing a local representative is unavoidable in any event, irrespective of whether national law formally requires it. A second group requires translation of the claims only (including Albania, Croatia, Denmark, Finland, Hungary, Iceland, Latvia, Lithuania, Montenegro, the Netherlands, North Macedonia, Norway, Slovenia and Sweden); there too, filing through a local representative is the norm.

Art. 4 of the Montenegrin Law on Patents illustrates how finely national law can draw these lines. Foreign persons with no seat or residence in Montenegro must in principle be represented by a registered representative or a Montenegrin attorney (para. 1). Expressly permitted without a representative, however, are filing the application, submitting priority documents, paying fees, and receiving the associated communications (para. 2) — provided an address for the receipt of documents in Montenegro is supplied (para. 3). If neither is provided, the office issues an invitation to remedy within three months and otherwise rejects the submission (paras. 4 and 5). Para. 6 adds that the maintenance fee may be paid by any person in the name of the proprietor. Paying the renewal fee in Montenegro therefore requires no representative — but it does require an address for service.

Stage 3 — address for service

Even in those states where the European patent takes effect without translation and without any formal step — Belgium, France, Germany, Ireland, Luxembourg, Malta, Monaco, Switzerland/Liechtenstein and the United Kingdom — appointing a local address for service is advisable. Without one, communications from the national office may well not reach the proprietor, and that includes the reminder about an outstanding renewal fee or service of a revocation action. For Malta an address for service is effectively indispensable for proprietors from outside the EU; Switzerland and Liechtenstein require proof of an address in one of the two states.

Why this distinction matters: compulsory representation is not a mere formality but a liability issue. Where the domestic representative is missing, the proprietor cannot assert rights under the patent pursuant to § 25 PatG — the patent exists but is temporarily unenforceable. And where the address for service is missing, the national office’s fee reminder goes nowhere: the six-month period runs out without the proprietor ever learning of it.

Takeaways for practice

  • Before grant, Rule 51(1) EPC applies uniformly — due on the last day of the month of filing.
  • After grant that uniformity fragments. 24 contracting states stay with the end of the month, 13 work to the anniversary, and Cyprus to the day before it.
  • The safe working rule is to docket to the anniversary of the filing date throughout after grant, and separately to the preceding day for Cyprus.
  • “Due on the filing date” has no place in a docketing instruction. What is meant is the anniversary of the filing date; the confusion generates systematic errors once entries are processed automatically.
  • Compulsory representation is a three-stage question: Art. 133(2) EPC for EPO proceedings, national law after validation (in Germany § 25 PatG), and an address for service even in states with no validation formalities.
Posted in European Law, International Law, Patent Law | Tagged , , , | Leave a comment

The “Black Hole”: Why the German Priority Document Remains a Legal Risk at the EPO

Posted on July 22, 2026 by Dr. Dirk Franke

Anyone filing a European patent application that claims a German priority must supply the priority document — yet receives no confirmation from the EPO that it has arrived and been accepted. Since the DPMA joined WIPO DAS there has been an elegant way out. But it is neither automatic nor forgiving of mistakes. A critical assessment — with concrete practical guidance.

Among the inconspicuous but dangerous formalities in European patent proceedings is the priority document. Under Rule 53(1) EPC, an applicant claiming priority must file a certified copy of the previous application within 16 months of the earliest priority date claimed. If the document is not filed in time, the right of priority is at risk (Rule 59, Article 90(5) EPC). This sounds like routine — but it is not, once the document disappears into a “black hole” at the EPO.

No receipt, no certainty

The EPO does not positively confirm the substantive receipt of a subsequently filed priority document. Electronic filing does generate an automatic acknowledgement of receipt for the transmission of the files; however, there is no communication confirming that the document has actually been placed in the file and recognised as compliant with Rule 53. The applicant learns of a problem — if at all — only through an invitation under Rule 59 EPC, and that invitation may come late. In the worst case, the deficiency surfaces only during examination or even in opposition, when a cure is long since impossible and the priority — and with it, potentially, the validity of the patent — has been lost.

Progress with a catch: the DPMA joins WIPO DAS

For a long time this risk could not be avoided elegantly for German priorities, because German priority documents could not be exchanged electronically; the certified paper copy was mandatory. That has changed: since 25 November 2024, the DPMA participates in the WIPO Digital Access Service (DAS) as a “depositing Office” for patents and utility models. A German priority document can now be made available electronically in DAS and retrieved by the EPO — free of charge and without paper (whereas the certified paper copy is subject to a fee).

The catch lies in the procedure. The exchange is not automatic but request-based: the applicant must actively request the DPMA to make an electronic priority document available (form A 9164) and receives a confidential access code. That code must then be communicated to the second office — here the EPO — so that the EPO can retrieve the document from DAS. If either step is omitted, the document is not in DAS, and the certified paper copy — with the confirmation gap described above — remains the only option.

Why Germany remains a special case

To make matters worse, the EPO does not retrieve German priority documents of its own motion. For a number of offices the EPO obtains the document ex officio and without any action by the applicant — for European and EPO-filed PCT first filings, and, on the basis of established exchange arrangements, for Chinese, Korean and US first filings (Rule 53(2) EPC; Guidelines A-III, 6.7). Germany is not on this list for automatic retrieval. For German priorities, therefore, the applicant must always supply the DAS access code. As a result, a German priority is treated, in formal terms, more awkwardly at the EPO than a Chinese or US one — a remarkable state of affairs for the home country of many EPO applicants.

The legal consequence — and why you cannot rely on the safety net

If the document is missing, the EPO invites the applicant under Rule 59 EPC to file it within a specified (in practice two-month) period. If the applicant fails to respond, the right of priority is lost (Article 90(5) EPC). That invitation, however, is no reliable lifeline: its timing is not guaranteed, it may issue late, and a subsequent cure is conceivable — if at all — only under the strict conditions of re-establishment of rights (Article 122 EPC). Anyone who makes the securing of priority depend on an official reminder shifts an avoidable risk onto an uncertain automatism.

How to meet the deadline with legal certainty

In practice, the “black hole” can be defused. The following approach is advisable:

  • Choose the DAS route deliberately and early: already at the German first filing, request the electronic priority document (DPMA form A 9164), document the access code and provide it to the EPO (e.g. in the Request for Grant, Form 1001). Retrieval by the EPO is then free of charge and on the record.
  • With a paper document — early and with proof: if a certified copy is filed after all, do so via electronic filing well before the 16-month deadline expires; keep the automatic acknowledgement of receipt as evidence of transmission.
  • In important cases, use belt and braces: provide the DAS access code and additionally file a certified copy — the minor extra effort is out of all proportion to the threatened loss of priority.
  • Verify actively instead of trusting: after filing, check in the European Patent Register (file inspection, “All documents”) and in MyEPO whether the priority document is actually in the file or has been retrieved. Do not wait for a confirmation — none will come.
  • Monitor the deadline yourself: docket the 16-month deadline under Rule 53(1) EPC independently and work with a buffer; the Rule 59 invitation is no substitute for your own deadline management.
  • Add the DAS request to the first-filing checklist: because the DPMA deposit is request-based, “request electronic priority document” belongs firmly in the workflow of every German first filing from which a priority claim may later be built.

Conclusion

The DPMA’s accession to WIPO DAS is genuine progress — but it removes the legal uncertainty only for those who operate the procedure actively and correctly. As long as the EPO does not retrieve German documents automatically and issues no positive proof of receipt for the priority document, securing priority remains the applicant’s own responsibility. The good news: with the DAS access code, deliberate double protection and a checking glance at the Register, the “black hole” can be closed reliably — one simply has to do it consistently.

Posted in European Law, Patent Law | Tagged , , | Leave a comment

Pressure to Register and Open Data Floodgates: WIPO’s Questionable Communication Practice

Posted on July 22, 2026 by Dr. Dirk Franke

Anyone filing an international patent application (PCT) today is soon pushed to register in WIPO’s system – and finds their inbox filling up with dubious payment demands. Neither is a coincidence; both are the result of a policy that consistently shifts burdens and risks onto applicants. A critical assessment.

Anyone who files an international patent application under the Patent Cooperation Treaty (PCT) – for example through the European Patent Office (EPO) as receiving Office – regularly experiences two things. Shortly after filing, messages arrive from the World Intellectual Property Organization (WIPO) urging registration in its electronic ePCT system. And the e-mail address provided at filing fills up within weeks with deceptively authentic-looking “invoices” and “registration offers”. Both phenomena deserve a critical look.

From Optional Convenience to De Facto Compulsion

As recently as 2008, electronic communication was a mere add-on service: by ticking a box on the PCT request form, the applicant could authorise an “advance copy” of notifications by e-mail. The legally decisive channel remained delivery on paper. WIPO put it unmistakably at the time: the e-mail communications did not replace the “paper notifications sent by mail as usual, which will remain the legal copy”.

This coexistence ended abruptly. On 30 March 2020, in the course of its COVID-19 business-continuity protocol, WIPO’s International Bureau stopped sending paper documents and announced that it would henceforth transmit PCT documents “exclusively via email”. What began as a temporary pandemic measure has to this day – in 2026 – never been reversed. Electronic delivery has thus shifted from a convenience to the de facto standard; the traditional paper mail that many applicants prefer for good reason has practically ceased to exist. A voluntary add-on option has become – without any formal change in the law, solely through a never-ended “emergency measure” – a largely alternative-less channel.

The Pressure to Register: Account, App and “eHandshake”

In parallel, WIPO pushes its users into the ePCT system. Anyone who wants to view an application online, actively manage it or make submissions needs a “WIPO account” with “strong authentication” – in practice a smartphone app or a security token. Shared access to a file is additionally controlled via so-called “eHandshakes”, that is, the mutual linking of authenticated WIPO accounts before any access rights (eOwner, eEditor, eViewer) can even be assigned.

For large firms with a steady PCT practice this may be manageable. For occasional applicants, small and medium-sized enterprises or individual inventors, however, it means that in order merely to receive official communications they are expected to familiarise themselves with an account, app and rights-management system. Many – understandably – do not want to. Particularly unfortunate here is an inherent contradiction: the very organisation that sends unsolicited prompts to register and confirm an account warns elsewhere, in the strongest terms, against precisely such “official-looking” e-mails. For the recipient, the line between a legitimate prompt and phishing becomes blurred – a home-made problem.

The Open Data Source: How a Filing Turns Into Spam

The second point of criticism weighs more heavily. The International Bureau publishes every PCT application 18 months after the priority date in the freely accessible PATENTSCOPE database – with the names and addresses of applicants and representatives, application and publication numbers, title, IPC symbols and priority details. This data is systematically harvested and used to fuel a decades-old industry of misleading requests for payment: deceptively authentic-looking “invoices” for supposed registrations or publications that have nothing to do with WIPO.

The scale is remarkable. WIPO itself maintains a warning list of fraudulent senders covering the period from 2002 to 2024 – a documented problem spanning more than twenty years. Recently the attack vector has shifted: alongside the classic paper “invoices”, e-mail phishing is on the rise. In a warning dated 10 April 2026, WIPO reports fake e-mails impersonating WIPO, the EPO and the EUIPO – for instance via spoofed addresses such as admin@wipo-office.com – demanding payment for the supposed securing of IP rights. WIPO’s key statement reads: “None of these organizations – WIPO, EPO or EUIPO – will solicit payments through unsolicited emails”. Anyone filing a PCT application therefore feeds their contact data into an environment whose misuse the organisation has long been aware of.

In Fairness: The Spam Does Not Come From WIPO – But WIPO Enables It

At this point, fairness is called for. The dubious invoices and phishing e-mails do not come from WIPO. They are third-party fraud, against which WIPO expressly warns and which it occasionally even helps to pursue – as in the case stopped by the Florida Attorney General in 2009. Anyone who blames the flood of spam directly on WIPO misreads the facts.

And yet the criticism remains justified. For WIPO is not a neutral bystander but the operator of the infrastructure that makes the abuse possible in the first place. It obliges applicants to disclose their contact data, keeps the data source fully open year after year, and shifts the entire burden of vigilance onto those affected by way of a warning notice. At the same time, by abolishing paper delivery it has removed precisely that low-threshold, hard-to-scale channel that many applicants valued as a robust fallback. The result is a structure in which applicants must either enter WIPO’s ecosystem or risk missing official communications – and in which their data flows, in any event, into a fraud market known since 2002. An organisation that has documented the problem for so long could do more than warn: it could refrain from publicly exposing e-mail addresses, offer a genuine opt-out or paper option, and act more decisively against the senders.

What Applicants Can Do

Until the practice changes, a structured approach is advisable:

  • Consolidate the address for service: Where possible, route official communications through the professional representative (as “common representative”) rather than exposing your own company or private address to direct delivery.
  • Use a dedicated e-mail address: For IP matters, use a separate address or alias that can easily be replaced if abused.
  • Do not pay unsolicited invoices: Check the sender address and the payment channel. Genuine WIPO communications end in @wipo.int, and neither WIPO nor the EPO demands payment by unsolicited e-mail. Only the International Bureau charges fees for international publication – there is no separate third-party “publication fee”.
  • Establish internal approvals: Clearly define who in the organisation is authorised to approve payments for IP rights.
  • Monitor deadlines independently: Do not rely on WIPO e-mails arriving reliably; note and monitor deadlines yourself.

Conclusion

WIPO’s communication practice is convenient for the organisation and inconvenient for applicants. The quietly perpetuated farewell to paper, the pressure to register and the unrestricted publication of contact data together paint a picture that is hard to reconcile with the claim of being a reliable, applicant-friendly authority. Until WIPO improves matters here, the rule is: every “official-looking” message following a PCT filing deserves a healthy dose of scepticism – and a firm processing protocol within the firm.

Posted in International Law, Patent Law | Tagged , , | Leave a comment

Patent Protection in the Netherlands: The Registration Patent, the PCT Gap, and a Reform on the Horizon

Posted on July 10, 2026 by Dr. Dirk Franke

The Netherlands occupies a special position in European patent law that many applicants only discover once they actually have to work with it: a Dutch national patent is, to this day, granted without any substantive examination of novelty or inventive step — a pure “registration patent”. At the same time, a PCT application currently cannot be converted directly into the Dutch national phase. Both points are the subject of a far-reaching reform that, at the time of writing, is still working its way through the legislative process. The overview below sets out the current legal position and the changes on the horizon.

1. Direct Filing: A National Patent Application in the Netherlands

A direct, stand-alone national patent application with the Dutch patent office — Octrooicentrum Nederland, part of the Netherlands Enterprise Agency (RVO), based in The Hague — is possible at any time, whether as a first filing or as a subsequent filing claiming a foreign priority under the Paris Convention. This route is entirely independent of the PCT issue described below and remains the classic, unproblematic path to Dutch patent protection.

Also common in practice is the so-called “NL-PCT strategy”: a Dutch priority application is filed first at the national level; within the twelve-month priority period, an international PCT follow-up application is then filed, for which the EPO — as the competent International Searching Authority — has by then already produced a novelty report on the original Dutch application. This gives the applicant roughly two and a half years to make the eventual choice of countries based on a solid search report. This is purely a filing strategy and has nothing to do with the separate question of entering the Netherlands from a PCT application, addressed next.

2. Entering the Dutch National Phase from a PCT Application

This is where the most practically important particularity lies. Under the WIPO PCT Applicant’s Guide (in the version of the national chapter for the Netherlands current at the time of writing), the entry states explicitly: “The Office closed the national route” — the Office has closed the national phase for PCT applications. An international PCT application therefore cannot currently be converted directly into a Dutch national phase. The only competent designated/elected Office for the Netherlands under the PCT is the European Patent Office; via the PCT, only a European patent can be obtained for protection in the Netherlands, which is subsequently validated there.

Anyone seeking patent protection in the Netherlands via the PCT route must therefore go through the European phase before the EPO — a stand-alone Dutch follow-up filing derived directly from a PCT application is, under current law, not available. This is one of the central changes the legislative reform described below is intended to remove.

3. Formal Requirements

A Dutch direct filing requires a description, claims, drawings where applicable, and an abstract; an official novelty search is mandatory and, under a cooperation arrangement, is generally carried out by the EPO. Applicants without a residence or seat in the Netherlands must appoint a registered Dutch patent attorney (octrooigemachtigde), listed in the Register of Patent Attorneys (Octrooigemachtigdenregister), for the conduct of proceedings; this is not strictly required merely to secure a filing date, but it is required for further prosecution. As elsewhere in Europe, the term of protection is 20 years from the filing date, subject to payment of renewal fees.

4. Language

The description of a Dutch patent application may be filed in either Dutch or English — a useful cost saving where a European or international follow-up application in an EPO procedural language is planned anyway. The claims, by contrast, must be in Dutch; if they are initially filed in another language, a Dutch translation must be filed within three months of being requested to do so. The language of proceedings before the Office is otherwise Dutch.

5. Particularities of Dutch Patent Law

The registration patent. The substantive patentability requirements under the Rijksoctrooiwet 1995 (novelty, inventive step, industrial applicability) are harmonised with the EPC. They are not, however, examined by the Office of its own motion before grant: the Office merely produces a novelty search report, but grants the patent regardless of its outcome. Whether an invention was, in fact, novel and inventive is only established if the patent is challenged in court. This makes Dutch patents quick and inexpensive to obtain, but legally less robust than examined patents — a trade-off that has been the subject of criticism for years.

The advisory opinion procedure (Art. 76 ROW 1995). Anyone wishing to argue the invalidity of a Dutch patent in court must submit to the court an advisory opinion from Octrooicentrum Nederland on the grounds of invalidity. This opinion is not binding, but in practice provides an initial, technically grounded assessment of validity — often the basis for an out-of-court settlement.

The prohibition of double protection vis-à-vis the European patent (Art. 77 ROW 1995). Where both a Dutch national patent and a European patent are granted for the same invention with the same filing or priority date, the Dutch national patent loses its effect once the European patent becomes unassailable. If the European patent is later (partially) revoked, the national patent does not revive. In practice, this means that building up parallel protection through both routes is not an additional safeguard — it results in the loss of the national right.

No opposition procedure. There is no opposition procedure comparable to that before the EPO; review of validity takes place exclusively before the ordinary courts, supplemented by the advisory opinion procedure described above.

Unitary Patent and the UPC. The Netherlands is a contracting state to the Agreement on a Unified Patent Court and has participated in the Unitary Patent system from the outset. A Unitary Patent, or a classic European patent litigated before the UPC, accordingly takes effect in the Netherlands without separate validation — a contrast, for instance, with the Czech Republic or Poland, which (as of this article) have not yet ratified the UPC Agreement.

6. The Reform on the Horizon: the Nieuwe Rijksoctrooiwet (NROW)

In early 2026, the Dutch government agreed on a fundamental reform of patent law. Its legislative process is, at the time of writing, still underway and it is accordingly not yet in force. The central proposed changes are:

  • Reintroduction of the examined patent. In future, a Dutch patent is to be granted only once novelty, inventive step and sufficiency of disclosure have been officially examined — the current pure registration system would end for new applications. This would considerably increase the legal certainty of granted patents, but is likely, in return, to lengthen and raise the cost of examination.
  • Opening up PCT national-phase entry. International applicants would in future be able to enter the Dutch national phase directly from a PCT application, without having to take the detour via the European phase before the EPO — precisely the gap described in point 2 above.
  • Protection for the North Sea economic zone. Patent protection is to be explicitly extended to the exclusive economic zone in the North Sea — relevant, for example, to offshore wind energy and cable and pipeline infrastructure.
  • Easier third-party objections against patents considered by third parties to have been wrongly granted, as a kind of corrective for the disappearing pure registration principle.

As the legislative process is still in parliamentary deliberation, neither the exact date of entry into force nor the final shape of the transitional provisions for pending applications has been settled. Applicants with pending Dutch proceedings should keep track of further developments, since it may become possible to request examination voluntarily for applications already pending.

Conclusion for IP Practice

  • Direct filing: unrestricted. A national Dutch patent application — whether a first filing or a subsequent filing — can be filed with Octrooicentrum Nederland at any time.
  • PCT national-phase entry: currently unavailable. Anyone seeking protection in the Netherlands via the PCT must go through the European phase before the EPO; a direct Dutch national phase does not currently exist.
  • Language flexible, claims mandatorily in Dutch. The description may be filed in English; the claims must be in Dutch (at the latest, within three months of being requested).
  • Registration patent, not examined patent — for now. Dutch patents are, to date, granted without substantive examination, making them fast and inexpensive but legally less robust than examined rights.
  • Mind the double-protection bar. Parallel protection through a national and a European patent for the same invention results in the loss of the national patent once the European patent becomes unassailable — it is not an additional safeguard.
  • Reform is coming. The planned Nieuwe Rijksoctrooiwet would reintroduce examination and close the PCT national-phase gap, but remains in the legislative process for now.

Photo: © Nicolas Raymond, [CC BY 2.0]

Posted in European Law, Patent Law | Tagged , | Leave a comment

Can Relevant Prior Art Be “Hidden” Behind the Closest Prior Art?

Posted on July 9, 2026 by Dr. Dirk Franke
European Patent Office

Under EPC patent law, can more relevant — in particular, technically closer — prior art be “hidden” by instead relying on a different document with a similar purpose as the closest prior art (CPA)? The question is sharpened by the suspicion that the skilled person might somehow become “less skilled” if the more relevant document was published later and is therefore not considered as a starting point. The short answer is: No. Neither the chronological order of prior publications nor the choice of a document with a similar purpose can shield relevant prior art from examination. At the same time, the question does expose a genuine weak point of the approach — the imprecision inherent in the selection stage.

1. The Skilled Person Knows the Entire State of the Art

The skilled person under the EPC is not a real-world searcher who finds some documents and overlooks others, but a legal fiction. This person is deemed to have had access to the entire state of the art under Art. 54(2) EPC. The publication date determines only whether a document forms part of the prior art at all (publication before the filing or priority date) — not whether it is “findable”. Two documents published before the relevant date are equally available, regardless of which one appeared first. “Hiding” relevant prior art through later (but still pre-published) publication is therefore excluded from the outset.

2. Selecting the Closest Prior Art

Under the Guidelines for Examination, G-VII, 5.1, the closest prior art is a document directed to a similar purpose or effect — or at least belonging to the same or a closely related technical field — that requires the fewest structural and functional modifications. This purpose-based criterion is the hindsight-free criterion: the purpose of a document can be determined independently of the claimed solution. Selecting the CPA based merely on the number of features shared with the claim, by contrast, would itself be tainted by hindsight, since it presupposes knowledge of the claim.

3. Multiple Realistic Starting Points

Under settled case law (T 967/97, T 21/08, confirmed in T 1742/12), where the skilled person has several viable routes starting from different documents, inventive step must be assessed against all of these routes. If the invention is obvious from even a single realistic route, inventive step is lacking. Two consequences follow: a document with a less similar purpose is not discarded merely because the document with the more similar purpose failed to lead to the invention; and, conversely, a finding of non-obviousness starting from the purpose-related document does not save the invention if it is obvious from another realistic starting point.

T 1742/12 frames this as a logical inversion: a document from which the invention is obvious is, by definition, the “more promising springboard” compared with one from which it is not. And T 405/14 makes clear that the closest prior art need not be the technically closest document: a document sharing the same purpose and many features often does not support a convincing obviousness objection, whereas the invention may follow, without hindsight, as obvious from an apparently less promising document.

4. Hindsight in the Selection Stage — and Where It Is Actually Controlled

The objection that choosing a less purpose-related CPA is itself an act of hindsight is, in part, correct. In T 855/15, the Board held that the question of whether the skilled person “would” select a document in order to arrive at the claimed invention is itself hindsight. A results-oriented selection of the springboard is therefore impermissible. That said, the mere technical distance of a document does not, on its own, exclude its use as a starting point.

Hindsight is accordingly not controlled by prohibiting distant starting points, but at two other stages of the analysis:

  • The “could-would” approach: The path from the starting point to the invention must show that the skilled person actually would have made the modification without knowledge of the solution — not merely could have.
  • A hindsight-free technical problem (T 605/20): The objective technical problem must not be formulated using the solution itself. Where the problem is tailored so as to already point toward the solution, this is precisely where the prohibited hindsight lies.

In practice, distant starting points typically fail at this second stage: they require more modifications, and for each one, a hindsight-free “would” — rather than merely “could” — must be established.

5. The Procedural Asymmetry

To reject a claim, one realistic starting point suffices, and under T 967/97 its selection need not be specifically justified. To uphold inventive step, by contrast, the invention must withstand analysis from all realistic routes. This asymmetry means that any imprecision in the selection stage tends to work in the attacker’s favour — a key reason why the doctrine remains controversial.

6. Criticism: The Imprecision of the Selection Stage

Terms such as “realistic”, “similar purpose” and “neighbouring field” have no objectively sharp boundaries. Since a chain of reasoning cannot be sharper than its least precise link, the overall assessment inherits this indeterminacy — even where the later steps are more disciplined. Parts of the academic literature accordingly describe “closest prior art” as a misleading concept.

The EPO’s defence is not that the concept is, after all, objective, but that its imprecision is rendered harmless: the multiple-starting-points doctrine dissolves the ranking question (“is A or B closer?”) because it never needs to be answered in the first place (T 1742/12: suitability rather than proximity). Moreover, the imprecision sits only at the margin — with borderline “realistic” documents — and this margin is caught in any event by the could-would stage. Finally, legal terms such as “similar” or “reasonable” are standards, not algorithms; their binding force arises from the duty to give reasons, from case-by-case development, and from appellate review. “Not objectively sharp” therefore means bounded discretion, not arbitrariness.

The honest assessment lies in between: the method is not fully objective, but nor is it arbitrary. It is a structured value judgment — and just how tightly that structure binds remains a genuine point of dispute.

7. Contrast with the German Federal Court of Justice (BGH)

The German Federal Court of Justice (Bundesgerichtshof) deliberately takes a less schematic approach. It does not require a search for the “closest” prior art, but rather a comprehensibly reasoned, purpose-oriented selection decision by the skilled person. This does not eliminate subjectivity, but it does raise the burden of reasoning for the choice of starting point.

8. Conclusion

Relevant prior art cannot be “hidden” either by later publication or by the choice of a purpose-related CPA: the fictional skilled person knows everything, and any more relevant document can itself serve as a (further) realistic starting point for an obviousness attack. The legitimate core of the objection does not concern “hiding” as such, but rather the imprecision of the selection stage — which the EPO does not control at the selection stage itself, but only at the subsequent, hindsight-free obviousness analysis.

Overview: Relevant EPO Decisions

Case numberKeywordCore holding
T 967/97Multiple starting points / no justification required for choiceWhere several viable routes exist, inventive step must be assessed against all of them; no special justification is required for the choice of starting point when denying inventive step. What matters is suitability, not “proximity”.
T 21/08All realistic routesConfirms: if the invention is obvious from even a single realistic route, inventive step is lacking.
T 1742/12Most promising springboardA document from which the invention is obvious is, by definition, the more promising springboard; a different purpose does not bar the analysis (with T 824/05).
T 405/14Closest ≠ technically nearestA document with the same purpose and many shared features often does not support a convincing obviousness objection, whereas an apparently less promising document may lead to the invention without hindsight.
T 1841/11Similar purpose sufficesA document with a similar purpose is not disqualified as CPA merely because another document with the same purpose exists.
T 855/15Hindsight in the selection stageAsking whether the skilled person “would” select a document in order to arrive at the claimed invention is itself hindsight. Mere distance, however, does not exclude the document from consideration.
T 605/20Hindsight-free technical problemThe objective technical problem must not be formulated using the solution.

Legal basis: Articles 54(2), 56 EPC; Guidelines for Examination, G-VII, 5.1. Further decision cited: T 824/05 (with T 1742/12).

Posted in European Law, Patent Law | Tagged , , | Leave a comment

The Quiet Cost Creep: How the EPO Has Multiplied Its Fees, Step by Step, in Recent Years

Posted on July 7, 2026 by Dr. Dirk Franke
European Patent Office

When the European Patent Office (EPO) announces a fee increase, the headline number is usually reassuring: “an average of 4%”, “around 5%”. These percentages are technically correct — and still misleading. They describe the mean across a fee schedule with well over a hundred line items, most of which any given applicant only ever pays a fraction of. Anyone who actually goes through a typical European patent procedure — a double-digit claim count, a PCT phase with a non-European search authority, the occasional divisional application, and ten or more years of renewal payments — experiences a very different cost trajectory. Alongside the regular, well-publicised fee rounds, the EPO has over the last decade and a half made a series of structural changes that individually made few headlines but add up to a substantial sum. Five of them are worth a closer look.

1. The Vanished Discount for Euro-PCT Applications with a Non-European Search Authority

Anyone who files an international application (PCT) with the US Patent and Trademark Office (USPTO), the Japanese Patent Office (JPO), the Korean Intellectual Property Office (KIPO), the Chinese patent office (CNIPA) or the Russian patent office as the International Searching Authority (ISA), and subsequently enters the European phase, must pay the EPO for a supplementary European search — after all, the EPO did not carry out the original international search itself. Until 2018, this supplementary search fee was reduced by a fixed amount for exactly this scenario: the reduction had been introduced in 2005 by decision of the Administrative Council (CA/D 10/05) and applied to applications whose international search had been carried out by the USPTO, JPO, KIPO, CNIPA, the Russian office (Rospatent) or the Australian patent office.

By decision of 13 December 2017 (CA/D 16/17, published in OJ EPO 2018, A3), the Administrative Council abolished this reduction outright as of 1 April 2018, with no replacement. Since then, applicants whose PCT search was carried out by one of these major non-European offices pay the full supplementary search fee — currently €1,595 (as of April 2026). The reduction survives only for a small group of European partner offices: Austria, Finland, Spain, Sweden, Turkey, plus the Nordic Patent Institute and the Visegrad Patent Institute.

In practice, this means that entry into the European phase has become noticeably more expensive since 2018 for exactly the applicants who most often route their PCT applications through the USPTO, JPO, KIPO or CNIPA as ISA — namely US, Japanese, Korean and Chinese applicants — without this showing up in any of the EPO’s regular “fee increase” announcements. It is not a fee increase in the strict sense; it is the quiet removal of a discount, with the same effect on the final invoice.

2. New and Rising Fees for Divisional Applications

Since 1 April 2014, the EPO has charged an additional filing fee for divisional applications of the second and any subsequent generation (decision CA/D 15/13 of 16 October 2013, OJ EPO 2014, A22). First-generation divisionals are exempt; from the second generation onward, the additional fee rises progressively, becoming a flat amount from the fifth generation. The EPO’s stated purpose was to make “long sequences of divisional applications” — and the resulting extension of pendency — less attractive.

When introduced in 2014, the additional fees were €210 (2nd generation), €420 (3rd generation), €630 (4th generation) and €840 (5th generation and beyond). Today the same line items stand at €235, €480, €715 and €955 respectively — an increase of 12–14% since introduction, even though this fee was untouched by the headline 2024 and 2026 reforms.

The real cost driver for divisionals now lies elsewhere, however: every divisional application requires retroactive payment of all renewal fees from the third year of the original filing date onward. Since precisely those early renewal fees — as shown below — have risen the most since 2024, filing a divisional today is substantially more expensive than it was just two years ago, even without the divisional-specific fee itself having moved.

3. Claims Fees: From an Afterthought to a Fixed Cost Item

Before 2008, a high claim count at the EPO was almost inconsequential: each claim from the eleventh onward carried a modest fee of €45. On 1 April 2008, the Administrative Council raised this amount for the 16th and each subsequent claim to €200 — a fourfold jump in one step. Just a year later, on 1 April 2009, the second stage followed: a separate, much higher rate of €500 was introduced for the 51st and each subsequent claim, described by commentators at the time as “draconian”.

Since then, this rate too has climbed steadily: as of 1 April 2024, the fee for the 16th to 50th claim rose from €265 to €275, and the fee for the 51st and each subsequent claim from €660 to €685. On 1 April 2026, further increases brought these to €290 and €720 respectively. Taken together, that is a roughly 45% increase over the 2009 starting point — for an application with, say, 60 claims (not unusual in many technical fields), claims fees alone quickly add up to a four-figure sum.

4. The Two-Tier Appeal Fee: An Increase Dressed Up as a Discount

Until 2018, appeals before the EPO’s Boards of Appeal carried a single, uniform fee: €1,880, regardless of who filed the appeal. On 1 April 2018, the Administrative Council introduced a two-tier structure: for natural persons, SMEs, non-profit organisations, universities and public research organisations (the entities listed in Rule 6(4) and (5) EPC), the fee stayed at €1,880. For every other appellant — in practice, any company that does not qualify as an SME — the fee rose to €2,255.

On 1 April 2020, the next step followed: the standard fee jumped to €2,705, while the reduced fee edged up only modestly to €1,955. Today the rates stand at €2,925 (standard) and €2,015 (reduced) — figures that were left unchanged in 2024, and are accordingly reported in official communications as “no change”, even though they are the result of the two very substantial increases that preceded them.

Taken as a whole, the appeal fee for the majority of represented clients — companies that are not SMEs — has risen from €1,880 to €2,925 since 2018, an increase of around 56%. The “reduced” fee reserved for individual inventors, SMEs and universities grew by only about 7% over the same period (€1,880 to €2,015). The official narrative consistently emphasises the social dimension — protecting smaller applicants — while obscuring the fact that a substantial fee increase was simply introduced for the majority of appeals actually filed, packaged as the creation of a new discount category. Anyone reading only the “appeal fee unchanged” announcements from 2024 and 2026 would never see this increase at all.

5. The “Linearisation” of Renewal Fees — the Biggest Lever

By far the largest, and least “quiet”, intervention concerns the renewal fees for the third to tenth year. Historically, these fees did not rise evenly but featured a conspicuous jump at the sixth year, reflecting the shorter average pendency times of earlier decades. By decision of 14 December 2023 (CA/D 16/23), effective 1 April 2024, the Administrative Council “linearised” this curve — with the stated aim of offsetting the revenue shortfall caused by significantly shorter average time-to-grant. Shorter pendency means fewer years in which the EPO itself collects renewal fees (after grant, they flow to the national offices instead) — the Office’s response was to raise sharply the early renewal fees that are still paid to the EPO.

The table below shows the trajectory from the pre-April-2024 level to the most recent increase on 1 April 2026 (decision CA/D 9/25 of 11 December 2025, roughly +5% across the board):

Renewal feepre-04/2024from 04/2024from 04/2026Total increase
Year 3€530€690€725+37%
Year 4€660€845€885+34%
Year 5€925€1,000€1,050+14%
Year 6€1,180€1,155€1,215+3%
Year 7€1,305€1,310€1,375+5%
Year 8€1,440€1,465€1,540+7%
Year 9€1,570€1,620€1,700+8%
Years 10–20€1,775€1,775€1,865+5%

The distribution is telling: the third- and fourth-year renewal fees — which practically every applicant must pay, regardless of whether the application is ultimately granted, refused, or withdrawn — rose the most, by +37% and +34% respectively. The sixth-year fee was actually cut slightly in 2024, before rising again in 2026. For a procedure of normal length, the net effect is a substantial front-loading of cost into the earlier, less certain years of the procedure.

6. Further Increases in the Slipstream of the Reforms

The 2024 and 2026 reforms also raised the other core procedural fees, if more moderately:

  • Search fee: €1,460 → €1,520 (2024) → €1,595 (2026), a total of +9%
  • Examination fee: €1,840 → €1,915 → €2,010, +9%
  • Designation fee: €660 → €685 → €720, +9%
  • Grant fee: €1,040 → €1,080 → €1,135, +9%

By deliberate contrast, the filing fee, the opposition fee, and — since 2020 — the appeal fee were left unchanged, a fact the EPO is happy to highlight in its communications because it makes entry into the procedure look inexpensive. That framing, however, distracts from where the real cost dynamics play out — as shown above: in claims fees, in divisional applications, in the search fee for Euro-PCT cases with a non-European ISA, in the appeal fee for companies, and above all in the early renewal fees.

7. Worked Example: A Divisional Application Compared, 2012 vs. 2026

To see how these individual effects add up in practice, consider a concrete example. Assume a second-generation divisional application, filed online, with 18 claims (three claims above the fee-free threshold of 15) and a 55-page specification (20 pages above the fee-free threshold of 35 pages). Due are the filing, search, examination and designation fees, the excess claims and excess pages fees, the renewal fees for years 3 through 6, and finally the grant fee. Compare the fee schedule of 1 April 2012 (decision CA/D 6/11 of 27 October 2011) with the current schedule as of 1 April 2026:

Fee item2012Qty2012 total20262026 totalIncrease
Filing fee (online)€1151€115€135€135+17%
Additional fee, 2nd-generation divisional(did not exist)1€0€235€235new
Excess pages fee (20 pages over 35)€14/page20€280€17/page€340+21%
Excess claims fee (3 claims over 15)€225/claim3€675€290/claim€870+29%
Search fee€1,1651€1,165€1,595€1,595+37%
Designation fee€5551€555€720€720+30%
Examination fee€1,5551€1,555€2,010€2,010+29%
Renewal fee, year 3€4451€445€725€725+63%
Renewal fee, year 4€5551€555€885€885+59%
Renewal fee, year 5€7751€775€1,050€1,050+35%
Renewal fee, year 6€9951€995€1,215€1,215+22%
Grant fee€8751€875€1,135€1,135+30%
Total€7,990€10,915+37%

For exactly the same divisional application — same claim count, same page count, same procedural steps — the official fees captured here alone have risen from roughly €7,990 (2012) to roughly €10,915 (2026): an increase of €2,925, or 37%, in 14 years. The distribution is notable: the third- and fourth-year renewal fees rose by +63% and +59% respectively, far outpacing the examination or designation fees (each around +30%). And the fee for a second-generation divisional — which did not exist at all in 2012, having only been introduced in 2014 — adds a brand-new €235 cost item by 2026, with nothing about the underlying procedure having changed. Together, these two effects explain why the overall 37% increase runs well ahead of what any of the individually moderate-looking fee rounds would suggest on their own.

Conclusion for IP Practice

  • The headline understates reality. “4%” or “5%” as an average obscures the fact that individual line items — claims fees, early renewal fees — have risen by 30–45% since 2009 and 2024 respectively.
  • Not every cost increase is labelled an “increase”. The abolition of the reduced search fee for Euro-PCT applications with USPTO, JPO, KIPO, CNIPA or Rospatent as ISA (2018) appears in no fee table as an “increase”, yet it costs affected applicants real money.
  • Divisional applications are now doubly expensive: through the generation fee that has risen since 2014, and — more significantly — through the sharply increased third- and fourth-year renewal fees since 2024, which must be paid retroactively for every divisional.
  • A “discount” can mask an increase. The two-tier appeal fee introduced in 2018 has raised the standard fee for companies by around 56% to date (€1,880 to €2,925), even as official communications have reported “no change” since 2020.
  • The early years carry the main burden. The linearisation of renewal fees shifts cost to exactly the point where practically every applicant must pay, regardless of the outcome of the procedure.
  • The worked example confirms the finding. For an identical divisional application (18 claims, 20 pages over the free threshold, renewal fees for years 3–6), official fees have risen from roughly €7,990 (2012) to roughly €10,915 (2026) — a 37% increase for an unchanged procedure.
  • Filing strategy matters more than ever. Trimming claim counts before filing, choosing the PCT search authority deliberately, scrutinising divisional chains critically, and making payments ahead of the relevant cut-off date (1 April) wherever the time limits allow — these are the most effective levers against a fee burden that has grown considerably faster in recent years than the official percentages suggest.
Posted in European Law, Patent Law | Tagged , , , , | Leave a comment

When the Description Reads Along: G 1/24, AI-Assisted Drafting, and the Firm’s New Liability Risk

Posted on July 5, 2026 by Dr. Dirk Franke

For decades, the description was the quiet part of a patent application. The music played in the claims; the description supplied background, embodiments, and fallback positions. As long as the claims were clear on their own, the exact wording of the description had few consequences in examination or litigation. Two developments are now unsettling this division of labor — and, of all places, they converge at exactly the same point. Legally, the EPO’s Enlarged Board of Appeal, with G 1/24, has elevated the description to a permanent tool of interpretation. Technologically, artificial intelligence (AI) is increasingly taking over the drafting of that very text. For law firms, this creates a risk that is easily overlooked: the AI’s efficiency gain is concentrated exactly where the duty of care is rising the fastest.

What G 1/24 Actually Decided

With its decision of 18 June 2025, the Enlarged Board of Appeal put an end to a long-simmering divergence in EPO case law. The core holding: the description and drawings must always be consulted when interpreting the claims for the purpose of assessing patentability — not only once a claim, read on its own, appears unclear or ambiguous. The underlying case concerned the interpretation of a single term (“gathered sheet”) in a patent; the Board used it to settle a fundamental question about how Articles 69 and 84 EPC interact.

Two points matter for practice. First, the decision harmonizes EPO practice with that of the Unified Patent Court and the national courts. What is written in the description will henceforth be read alongside the claims according to the same basic principle in examination, opposition, and infringement proceedings alike. The description thus moves from the periphery to the center of claim interpretation. Second — and precision matters here — the Board holds that the description must always be consulted for interpretation; it does not hold that the description may override an otherwise clear claim at will. The practical effect is therefore not a blank check to reinterpret every claim from the description. But it does shift the weight considerably: every definition, every use of a term, every statement about “the invention” can now influence how the claims are interpreted.

This is reinforced by a further, still open issue. Referral G 1/25 raises the question whether the description must be adapted to the claims prior to grant. Both directions carry risk: leaving contradictions in place may dilute the interpretation; adapting the description risks impermissible extensions (Art. 123(2) EPC) and unintended shifts in the scope of protection. However G 1/25 is decided, the message is already clear: the description is no longer neutral background text but an effective tool of interpretation.

From Background Text to a Sharp Tool

This changes the character of drafting itself. In the past, the description could be written generously, redundantly, and with many variants — extra sentences rarely did harm. After G 1/24, the opposite is true: every sentence now carries a potential interpretive effect. A definition that is broader or narrower than intended now counts. An offhand statement about the purpose or advantage of “the invention” can support a narrowing interpretation. An embodiment that promises more than the claim actually covers can become a point of attack. And terms used one way in one place and differently elsewhere create exactly the kind of inconsistency a court will now have to resolve — not necessarily in the patentee’s favor.

The description has thus become an instrument that demands consistency, terminological discipline, and a precise awareness of the interpretive consequences of every word. These are exactly the qualities that AI-generated text systematically struggles with.

Where the AI-Drafted Description Tips Over

AI can produce descriptions in a fraction of the time previously required — fluent, extensive, plausible. That is tempting, because the description is the most text-heavy and seemingly most mechanical part of drafting. But it is precisely the typical weaknesses of generative models that strike at the risks G 1/24 has sharpened:

Terminological inconsistency. Language models aim for variation and fluent phrasing, not rigid term fidelity. A model will use the same term slightly differently in different places, or introduce synonyms that carry a subtly different meaning. In a world where the description is always consulted for interpretation, that is no longer a cosmetic flaw — it is an interpretive trap.

Unsupported assertions and pseudo-embodiments. AI tends to fill gaps “sensibly” — with advantages, mechanisms of action, or variants that are not technically supported. Such passages can steer claim interpretation in an unwanted direction, and — should G 1/25 confirm a duty to adapt — become a source of Article 123(2) problems when later deleted.

Scope bias. Generative models, when in doubt, produce more text, not less. But more text means more surface area for contradictions, overreaching definitions, and incidental commitments — in other words, more of exactly what a court now consults.

Imported phrasing patterns. AI draws on training data full of other parties’ patents. It adopts “characterizing” phrases, implicit disclaimers, or purpose-bound language that quietly shifts the scope of protection — phrasing an experienced drafter would have deliberately avoided.

Definitions with a life of their own. A model likes to write its own definitions into the description. If such a definition — narrower or broader — departs from the intended claim meaning, it now has a direct effect on interpretation under G 1/24.

The insidious part is that these errors don’t look like errors. AI text reads competently and fluently. The inconsistency is not hidden in an obvious blunder but in the interplay of passages that each look innocuous on their own — and only reveal their effect once an infringement dispute arises.

The Firm’s Liability Risk

For a law firm, this bundles into a specific risk. The patent attorney is responsible; they sign the application, and they are liable if an unnoticed inconsistency in the description narrows the client’s scope of protection or brings a patent down in litigation. That responsibility cannot be delegated to a model.

The real core of the issue is a paradox. AI saves the most time exactly where G 1/24 demands the greatest care. The description used to be the part one could most safely “let run through” — and it is now the part that demands the most precise, interpretation-aware scrutiny. The apparent efficiency gain is therefore deceptive: whoever merely skims the AI-drafted description hasn’t worked faster — they have simply pushed the risk, invisibly, into the future: into examination, opposition, and infringement proceedings.

On top of that, the necessary review is not the fast kind of review. Checking an AI-drafted description for interpretive consequences means reading the entire text against the claims: for consistent terminology, for definitions, for overreaching statements about “the invention,” for unsupported embodiments. That comparison is cognitively demanding and often no faster than drafting it oneself — sometimes slower, because someone else’s plausible-sounding phrasing first has to be deconstructed to recognize its interpretive effect. The AI’s time advantage shrinks in direct proportion to the care that G 1/24 now requires.

This is not an argument against AI. It is a precise mapping of its risk: the description is the place where AI’s characteristic weaknesses meet the stakes that G 1/24 has raised. This is exactly where it is decided whether a firm has mastered AI — or whether AI has written a liability problem into the file.

What Firms Should Do Now

The analysis points to concrete safeguards that preserve AI’s efficiency gain without increasing liability:

  • Terminology and definition review. Before filing, a targeted check: is every claim-relevant term used consistently and in its intended meaning throughout the description? AI-generated synonyms and self-made definitions must be explicitly reviewed.
  • Consistency check against the claims. The description must be read as an interpretive instrument, not as continuous prose. Every statement about purpose, advantage, or “the invention” must be examined for its potentially narrowing effect.
  • Restraint with boilerplate. What counts under G 1/24 should be intentional. Overreaching embodiments and unsupported assertions are more of a risk than a fallback position.
  • G 1/25 readiness. As long as the duty to adapt remains unresolved, the description should be drafted so that later conformity amendments create as little Article 123(2) exposure as possible.
  • AI as a draft, not a result. The sensible use case is an accelerated first draft followed by a documented, human interpretive review — not unreviewed adoption.
  • Process and documentation. A firm quality-control process — who performs the interpretive review and how it is recorded — protects both the client and the firm.

Conclusion for IP Practice

  • G 1/24 makes the description a permanent tool of interpretation. It is now always consulted to interpret the claims in examination, opposition, and — harmonized — before the UPC and the national courts. Every sentence now has an interpretive consequence.
  • AI text is weakest exactly where it now matters most. Terminological inconsistency, unsupported statements, scope bias, and imported phrasing patterns are exactly the errors G 1/24 makes dangerous — and they don’t look like errors.
  • The efficiency gain concentrates the liability risk. AI saves time precisely at the point that now demands the most careful, interpretation-aware review. A superficial check doesn’t save time — it merely shifts risk.
  • The review is not the fast kind of review. Checking the description against the claims is demanding and often no faster than drafting it oneself — the real time saved is smaller than it appears.
  • The firm’s value shifts toward interpretive skill. It is not the fastest drafting that wins, but the drafting that combines AI speed with human, litigation-aware judgment about every word.

G 1/24 thus provides an unexpectedly concrete legal basis for a claim that often stays abstract in the debate over AI in the patent profession: the human remains indispensable in the process — not as a matter of principle, but because the description has become a sharp tool that can only be safely wielded with judgment. Whoever uses AI as an accelerated draft and elevates the interpretive review to the core deliverable gains both: speed and safety. Whoever lets the description “run through unchecked” writes the risk into the file.

Posted in European Law, Patent Law | Tagged , , , , | Leave a comment

No Property, No Innovation: Why Socialist Patent Systems Forfeit Prosperity – The USSR, the GDR and Cuba

Posted on July 4, 2026 by Dr. Dirk Franke

A patent grants its holder the right to exclude others from using an invention. At first glance this looks anti-social — yet it is one of the most effective engines of prosperity that modern economies possess. It turns an idea into a tradable asset, makes research financeable, attracts capital, and rewards not only whoever invents a new process but whoever brings it to market. Socialist planned economies abolished precisely this right to exclude — internally. In doing so, this article argues, they sacrificed a central driver of innovation and economic power. The three cases of the Soviet Union, the GDR and Cuba show the mechanism and its price.

The economic starting point: property rights create prosperity — including intangible ones

That secure, enforceable property rights are the foundation of prosperity is among the best-supported findings in institutional economics. Douglass North showed that societies grow rich precisely where rules reliably assign the returns of productive effort to those who generate them; Daron Acemoglu and James Robinson traced differences in national wealth to “inclusive” versus “extractive” institutions, at the centre of which sits the property regime. Hernando de Soto coined the term “dead capital” for the absence of formal title: values that exist but cannot be mobilised as collateral, tradable goods, or a basis for investment.

What holds for land holds, in the knowledge economy, all the more for intangible assets. The value of modern companies today lies overwhelmingly in intangibles — patents, brands, know-how, data. A patent is the legal title that makes an invention capable of being booked, licensed, pledged and sold in the first place. Devalue that title and the invention does not disappear — but the incentive to invest in it, commercialise it and scale it does. This is exactly where the weakness of the socialist model begins.

1. The Soviet Union: invent much, exploit little

After the October Revolution, the Soviet Union nationalised inventions and created, as a socialist alternative to the patent, the inventor’s certificate (avtorskoe svidetel’stvo): the invention became state property, while the inventor received recognition and remuneration — but no right to exclude. Patents survived only as a parallel institution, used almost exclusively by foreigners. The logic behind it was deliberately anti-capitalist: knowledge was to flow freely between state enterprises rather than being locked up behind monopoly rights.

From the standpoint of institutional economics this was a consequential design flaw. The system optimised for the number of inventions and for the free flow of information — and blanked out the decisive variable: the incentive to actually turn an invention into a better product or process. Because enterprises operated as monopolies, were measured on plan fulfilment rather than competition, and the inventor’s reward was capped, the drive to introduce novelties was missing. The result is well documented in the literature as the “Soviet innovation problem”: much was invented, too little implemented and diffused. Observers attributed to Soviet patent law a dogmatic mindset that inhibited innovation rather than fostering it. Later Russian scholarship speaks of an institutional mismatch — of lost opportunities to generate growth from knowledge that already existed.

The contradiction is telling: the very institution meant to liberate knowledge decoupled inventing from exploiting — and thereby stripped the economy of the feedback loop that, in the West, turns inventions into productivity.

2. The GDR: the natural experiment

No other case illustrates the thesis as vividly as the German division — a controlled experiment with a shared language, culture and industrial history that differed only in economic system. East German patent law knew two routes: the exclusive patent (Ausschließungspatent), which corresponded to the Western patent, and the economic patent (Wirtschaftspatent), which could be used by all state-designated enterprises while the inventor received only a remuneration — capped, after the 1963 amendment, at 30,000 Mark. The state steered the choice forcefully: filing an economic patent cost 20 Mark, an exclusive patent 250 Mark, whose annual fees ran up to thirty times higher. For inventions arising in publicly owned enterprises, only the economic patent was permitted in any case.

The outcome was a patent order that effectively abolished the right to exclude: of roughly 111,000 GDR patents at reunification, only about 14,000 — just under 13 percent — were exclusive patents. A large body of protective rights thus existed with no tradable market value. The reaction, once a real market appeared, is revealing: more than 19,000 holders applied to convert their economic patent into an exclusive one. The right to exclude others became valuable at precisely the moment there were competitors and prices.

The economic record of this system is unambiguous. The GDR was regarded as the most developed country of the Eastern Bloc — and still lagged: by the late 1980s its per-capita GDP fell short of West Germany’s by around 30 percent, and labour productivity at unification stood at only about a third of the Western level. According to estimates by Akerlof and colleagues, only around ten percent of the GDR workforce was employed in firms viable at world-market prices; the East German economy collapsed in 1990/91 on a scale unprecedented in modern economic history. Research links this weakness directly to the centrally planned system, which systematically inhibited innovation, distorted the allocation of resources and undermined long-term growth. Even three decades later, and despite enormous transfers, the East reaches only around two-thirds of the Western per-capita level on many measures. Two systems, one people — and a productivity gap whose root lay substantially in the missing regime of incentives and property.

3. Cuba: the exception that proves the rule

Cuba appears to be the counter-argument: a socialist state that is a monopolist domestically and yet ranks among the notable innovators of international biotechnology. The state umbrella corporation BioCubaFarma reports around 2,640 patents in Cuba and worldwide, of which roughly 2,438 are held outside Cuba, and has registered 765 products in 53 countries. Developments such as Heberprot-P for diabetic foot ulcers, the cancer immunotherapy CIMAvax-EGF, or the COVID vaccines Abdala and Soberana are protected by patents in numerous jurisdictions.

On closer inspection, however, Cuba confirms the thesis rather than refuting it — for three reasons.

First, the success is concentrated in a narrow, state-selected niche that has been fed with prioritised resources over decades. It is the product of concentrated subsidy, not of a broad innovation culture nourished by ownership incentives. Outside this “scientific pole,” the Cuban economy is marked by chronic stagnation and by hard-currency and supply crises.

Second — and this is the decisive point — Cuba patents almost exclusively abroad. At home, where the state alone produces, the right to exclude is worthless; inventions are used freely within the country’s own health system. The patent becomes valuable only where there are markets, competitors and paying demand: in exports, in licensing, in joint ventures with foreign capital, for instance in the Mariel special development zone. To draw prosperity from its inventions, Cuba must therefore reach into precisely those property-based market economies that its own system denies internally. Here the patent is an export and hard-currency instrument — proof that innovation only creates value once it plugs into a property order.

Third, even this island runs into its limits: biotechnology is extremely capital-intensive, and capital is scarce in Cuba; the US embargo, international regulatory hurdles and a deep currency crisis constrain the returns. A state can finance world-class research at isolated points — but without the capital-forming power of tradable property rights, it cannot turn that into broad, self-sustaining economic strength.

The mechanism of failure

Bring the three cases together and a clear pattern emerges. Where the right to exclude intangibles is absent, several chains of causation collapse — the same chains that, in market economies, convert inventions into prosperity:

  • No price signal. Without exclusivity there is no market price for an invention — and therefore no information about which novelty is worth scarce resources.
  • No incentive to commercialise. Anyone who cannot exclude, and whose reward is capped, has little reason to undertake the arduous, risky work of implementation and scaling.
  • No capital formation. Intangible assets without legal title are “dead capital”: they cannot be pledged, traded or used as a basis for investment.
  • No market entry. Without company formation and competitors, the selective pressure that weeds out poor solutions and spreads good ones is missing.

The inventor’s certificate and the economic patent rewarded the inventor — but they severed the feedback between inventing and exploiting. Innovation became a cost item of the planning bureaucracy rather than a driver of productivity. The cumulative effect over decades is that very productivity gap which the German division made so relentlessly visible.

In fairness: not a plea for IP maximalism

This critique is no licence for a boundless patent system. The West, too, knows the downsides of excessive protection — patent thickets, “trolls,” blocked follow-on innovation, disputes over access to medicines. And Cuban biotechnology shows that targeted state research can produce real, humanitarically valuable results. The point is not “as much protection as possible,” but something more fundamental: an enforceable, tradable property right in intangible values is a necessary condition for turning inventions into lasting prosperity. That condition is exactly what socialist systems structurally failed to meet internally — and they paid for it in innovative capacity and economic performance.

Takeaways for IP practice

  • The right to exclude is not an end in itself but a prosperity mechanism. It makes inventions capable of being booked, financed and traded — the precondition for knowledge becoming value creation.
  • Reward without an incentive to exploit is not enough. The inventor’s certificate and the economic patent honoured the inventor without incentivising commercialisation — the GDR figures (only ~13% exclusive patents, then over 19,000 conversion requests) show where the market value really sat.
  • The German division is the litmus test. One people, two systems, a productivity gap of roughly two-thirds to one-third at unification — driven substantially by the missing regime of incentives and property.
  • Cuba confirms the rule. A monopolist patents by the thousand — but abroad, because value arises only where a property order sustains markets. Without capital formation out of intangibles, even world-class research remains an isolated island.
  • Intangible assets are today the core of corporate value. Whoever fails to secure them legally fails to mobilise them — and forfeits precisely the value creation that carries modern economies.

Socialist patent systems were no accident but the consistent application of an idea: that knowledge should belong to all. Economically, however, the uncomfortable finding holds that it was precisely the communalisation of the right to exclude that weakened the power to turn ideas into prosperity. Property — including, and especially, in intangible goods — is not an obstacle to innovation. It is its precondition.

Photo: © Nicolas Raymond, [CC BY 2.0]

Posted in International Law, Patent Law, Start-up Business Consulting | Tagged , , | Leave a comment

Who Survives AI? Partners, Associates and Freelance Patent Attorneys in a Shifting Firm Structure

Posted on July 4, 2026 by Dr. Dirk Franke

The debate about Artificial Intelligence (AI) in the patent profession usually focuses on the firm as a whole: falling drafting fees, consolidation pressure, and competition that hands the efficiency gains to clients. In our post “When the Patent Office Becomes More Expensive Than the Attorney” we developed that macro perspective. This piece zooms in — on the people a firm is actually made of. Because AI does not hit the three load-bearing roles of the patent profession evenly. It affects them very differently.

The starting structure: three roles, three risk profiles

Over decades, the patent profession settled into a clear division of labour:

Partners held the client relationships and the authority to decide who worked a case. They billed the work in their own name and carried the firm’s entrepreneurial risk — but they also held the one decisive resource: the relationship with the client.

Associates — employed or associated patent attorneys without equity — received work as it was allocated to them and billed it back to the partners largely without risk. They traded entrepreneurial risk for the security of a steady workload. As a rule, they had no direct client contact.

Freelance patent attorneys were brought in selectively: for workload peaks, for particularly demanding matters, or where they held specialist knowledge that could not be replaced. They bore the full risk of receiving no instructions, but in return they had to maintain no infrastructure. They, too, typically had no direct client contact.

This structure rested on one tacit premise: processing capacity is scarce and expensive. That is precisely the premise AI removes.

The single shift that explains everything

If the same work — claim drafting, the specification, prior-art searching, responding to office actions — can be done roughly five times faster, then the locus of value shifts. It moves away from execution and toward what machines (still) cannot do: client trust, strategic judgment, and personal accountability for the outcome.

From this follows a second, less obvious point — the question of who captures the efficiency gain. In the short term it is the partners: their costs fall while their prices to clients hold at the old level, and the margin widens. In the long term, however, competition passes the saving on to clients, because many attorneys can suddenly do far more work and compete for the same — not equally growing — demand. The transition from “short term” to “long term” is the real fault line, and each of the three groups experiences it differently.

1. Partners: short-term winners, under reform pressure in the long run

In the short term, partners are the clear beneficiaries. They own the one resource AI does not replicate — the client relationship — while their cost base falls: the same volume of work now requires fewer associate hours and less bought-in freelance capacity. As long as fees to clients still track the old levels, the margin expands.

In the long term, the picture reverses. Three forces bear down on partners:

First, price erosion. Competition transfers the efficiency gains to clients; revenue per matter falls. To hold their income, partners need either more matters or higher-value work.

Second, the loss of the infrastructure advantage. The classic lever of large firms was the pyramid: many associates whose hours the partner re-billed at a mark-up. Once AI takes over execution, this leverage shrinks — and with it the structural head-start that size once provided. A lean unit with AI can suddenly deliver quality that used to require a whole team.

Third, the threat of disintermediation. As soon as clients — especially large, patent-savvy corporations — deploy AI tools themselves, partners face a justification test: what exactly is the client still paying for? The answer can no longer be “capacity.”

Strategies for partners:

  • Move up the value chain. Portfolio strategy, freedom-to-operate analyses, licensing, opposition and infringement strategy, and translating imprecise invention disclosures into robust claims all remain judgment work. The partner as a provider of capacity loses; the partner as a strategic adviser and risk-bearer wins.
  • Reshape the pyramid into a diamond. Fewer pure-execution associates, more AI plus experienced judgment. Anyone who preserves the old leverage economics is funding structures the market no longer supports.
  • Productise the offering. Fixed-fee packages, subscription models and technology-enabled services replace the eroding hourly model for text production.
  • Own the AI layer. A firm that builds its own tools, prompts, review routines and quality assurance as a firm asset turns AI into an asset rather than a threat.
  • Open the market downward. Efficiency makes it possible to serve clients who previously failed on cost — SMEs, start-ups, individual inventors. If the price per matter falls, the number of matters can rise (a kind of Jevons effect for the patent profession). Falling total costs of a filing can generate additional demand — though this is capped by the rigid block of official fees, as described in the cost post.

2. Associates: the most exposed group

No role profile maps as precisely onto what AI does best as that of the associate: allocated execution work, billed to the partner. It is exactly this work that is being automated.

In the short term, the group splits in two. Those who master AI and become five times more productive themselves are more valuable than ever. Those who do not become dispensable. Hiring cools, because the leverage economics that once funded associate salaries yield fewer profitable hours.

A problem lurks here that reaches beyond the individual firm: the hollowed-out middle. Junior attorneys traditionally learn judgment on the simple work — the search, the first draft claim set, the routine office-action response. If AI takes over precisely this learning curve, the question becomes where tomorrow’s senior competence is supposed to come from. The profession risks cutting the training pipeline that feeds its future partners.

In the long term, the path to partnership narrows — fewer partners are needed, and the economics that funded that ascent are weaker. For the associate, two very different futures open up: the rise to AI-augmented senior who supervises, reviews and eventually holds relationships — or displacement toward freelancing. The associate who only executes has no future. The associate who builds judgment, client proximity, specialisation and AI mastery does.

Strategies for associates:

  • Become AI-fluent, fast. The decisive question is not whether AI does the work, but whether you are the one who does five times as much with it — or the one it replaces.
  • Break the “no client contact” rule. Building relationship and business-development skills early means acquiring exactly the resource that is scarce and valuable in the new environment.
  • Specialise deeply — in technology fields, procedures (such as opposition and nullity), or industries where judgment and experience command a premium and standardisation reaches its limits.
  • Occupy the “human in the loop” role. Quality assurance, sense-checking and responsibility for machine-generated drafts become a distinct, liability-relevant service in their own right.
  • Build a personal reputation that does not depend on the firm alone — through publications, talks, and visibility.
  • Think entrepreneurially. Because AI lowers the need for infrastructure, a capable associate can today set up a lean practice of their own with far less overhead than only a few years ago.

3. Freelance patent attorneys: structurally the biggest opportunity — and the biggest risk

Freelancers sit at the most interesting break point. Their classic reasons for being engaged — workload peaks and volume — partly dissolve, because partners can now absorb peaks with AI themselves. The “buffer for capacity spikes” is needed less.

In the short term, generalist freelancers therefore lose their overflow work. Specialists with irreplaceable knowledge keep their niche, at least initially.

At the same time, freelancers structurally have the best cost base of all: no infrastructure, no overhead, no administrative apparatus — and, with AI, enormous productivity. In a market of falling prices, the lowest-cost provider has the advantage. What they lack is precisely the most valuable thing: their own client access and a brand, the trust.

In the long term, that very question decides whether freelancers end up among the big winners or the big losers:

As winners, the free, AI-native patent attorney is the leanest conceivable unit — low fixed costs, high throughput, ideally positioned for a price-sensitive segment (start-ups, SMEs, individual inventors). They can significantly undercut established structures on price.

As losers, those who compete only as cheap execution capacity are finished — because there the AI itself is cheaper still. The pure commodity freelancer is ground down between the machine (all but free) and the firm (with its relationships).

Strategies for freelancers:

  • Occupy niches AI does not cover — cutting-edge technology, rare language or jurisdiction combinations, deep sector expertise, forensically demanding disputes.
  • Move from subcontractor to independent practitioner. AI removes the infrastructure barrier that once forced the freelancer into dependence on the firm. Building one’s own client relationships is the decisive step toward the micro-firm.
  • Network. Loose alliances and cooperatives can offer firm-like reliability without carrying the overhead of a classic firm.
  • Shift the offering — away from pure execution and toward AI-assisted senior review and second opinions that firms, too, will buy in.
  • Get ahead of disintermediation by the machine by offering precisely the responsibility, judgment and accountability that an AI cannot assume.

The common movement: all three roles converge on the same scarce competence

What is striking is that AI ultimately pushes all three groups in the same direction. The infrastructure advantage that separated the partner from the freelancer loses its significance. The execution capacity that associates and freelancers contributed becomes a commodity. What remains in the end and makes the difference is the same for all three: client trust, strategic judgment, and personal accountability.

For one constant endures: the patent attorney signs, takes responsibility, and is liable. That accountability cannot be delegated to a machine — it is the profession’s most durable human “moat.” Whoever combines it with genuine judgment and a resilient relationship survives the upheaval. Whoever sells only capacity will not.

Takeaways for IP practice

  • Your role, not your title, determines your exposure. Partners are short-term winners and under long-term reform pressure; associates are the most exposed; freelancers have both the greatest opportunity and the greatest risk.
  • Value migrates from execution to judgment. Each of the three groups should shift its time to where the machine cannot reach: strategy, the client relationship, responsibility.
  • AI competence is not an add-on but the entry ticket. The dividing line will run between those who do five times the work with AI and those it replaces.
  • The training question is strategic. If AI takes over the juniors’ learning curve, competence-building has to be reorganised — otherwise the profession will, a decade from now, lack the next generation of judgment.
  • The market can open downward. Falling costs can unlock new clients; the limiting factor remains the rigid block of official fees.

AI changes not only how much work is done in the patent profession, but who is paid for it and for what. The firm structure of partners, associates and freelancers was an answer to the scarcity of processing capacity. Once that scarcity is gone, the structure reorders itself — along the only resource that stays scarce: human trust and judgment.

Posted in Employee, Patent Law, Start-up Business Consulting | Tagged , , , | Leave a comment

When the Patent Office Costs More Than the Attorney: Two Scenarios for the Future of the IP Industry in the Age of Artificial Intelligence

Posted on July 2, 2026 by Dr. Dirk Franke

The Quiet Reversal of the Cost Structure

For decades, an unspoken rule of thumb governed the patent world: the office is cheap, the attorney is expensive. Anyone looking to reduce the cost of a patent filing turned the attorney screw — shorter specifications, leaner searches, more input from the inventor.

At the European Patent Office (EPO), that rule of thumb has long since ceased to apply. Official fees have overtaken attorney drafting fees. A European proceeding now costs more in pure official fees than the drafting of the application itself — and the gap is widening.

It is precisely this reversal that sets up the question set to reshape the IP industry in the coming years: What happens when Artificial Intelligence drastically reduces one side of the cost equation — and not the other?

The Cost Comparison in Numbers

Consider a typical constellation: entry into the European phase (or a direct filing) with 20 patent claims, pursued through the 5th renewal year. The figures are based on the EPO schedule of fees as in force from 1 April 2026.

EPO Official Fees

ItemAmount
Filing fee (online)€135
European search fee€1,595
Designation fee€720
Examination fee€2,010
Claims fees (claims 16–20: 5 × €290)€1,450
Subtotal — entry/examination≈ €5,910
3rd year renewal fee≈ €690
4th year renewal fee≈ €845
5th year renewal fee≈ €1,000
Subtotal — renewal fees (years 3–5)≈ €2,535
Total EPO official fees≈ €8,445

Note: Renewal fees at the application stage follow the applicable schedule of fees; the 3rd and 4th year renewals were raised above average with effect from 1 April 2026. The figures above are rounded to this order of magnitude.

Attorney Drafting

ItemAmount
Drafting and filing of the application€3,000 – €5,000

The result is striking and, for many clients, counterintuitive: even today — without any AI — the pure EPO official fees in this model case, at roughly €8,400, sit well above the attorney drafting fee of €3,000 to €5,000. The office has become the larger cost block, not the attorney.

This figure is the lever on which everything that follows turns. Because AI initially acts on only one of the two sides — the attorney’s.

Scenario 1: The Attorney Becomes Cheap, the Office Stays Rigid

AI-supported tools for claim drafting, specification generation, and prior-art search substantially reduce the attorney workload per application. The drafting fee — already the smaller item — falls further, to a fraction of its former level.

On the office side, in this scenario, nothing happens.

That is not an unrealistic assumption; it follows institutional logic. The EPO is a monopolist without competitive pressure. It pursues its own objectives, which are not aligned with the cost interests of applicants — among them safeguarding the employment and continued standing of a large examiner workforce. An authority with this incentive structure does not necessarily pass on AI-driven efficiency gains in the form of lower fees; it may just as readily channel them into institutional preservation, reserves, or political latitude. The most recent fee round — an increase in most core fees from 1 April 2026, with above-average rises in the early renewal fees — points to a rigid, upward-tending fee regime rather than to any pass-through of efficiency gains.

The consequences for firms:

If value creation per mandate shrinks on the attorney side, but overall demand for IP rights does not grow to the same degree, a displacement dynamic sets in. The same work is done by fewer people in less time. The number of mandates a single attorney can handle rises — the total number of mandates awarded does not.

The result is a market in which more and more attorneys compete for fewer and fewer economically viable mandates. The price of legal work falls, margins erode, and the efficiency gains from AI accrue not to the profession but are passed through to clients under competitive pressure. The office’s rigid cost block, meanwhile, prevents falling attorney costs from pushing the total cost of a filing low enough to noticeably stimulate demand. The office remains the bottleneck — and firms bear the adjustment burden alone.

In this scenario, AI is not a growth driver for firms but an accelerant of consolidation.

Scenario 2: Both Sides Give Way — The Filing-and-Grant Machine

The second scenario rests on a bolder assumption: the office, too, opens up to AI and passes on the efficiency gains. This becomes possible because here AI is not merely cheaper, but simultaneously better.

The decisive technical point is search. A near-perfect, machine-driven search for novelty-destroying prior art — across all languages, databases, and document classes — reduces the cost of the most labour-intensive part of examination while at the same time raising its quality. For the first time, cost reduction and quality improvement coincide rather than pulling against each other.

If both cost sides — attorney and office — fall at once, the entire economic calculus of the patent system tips over. Protecting even the smallest improvements becomes financeable. The number of new filings can rise dramatically.

At the end of this trajectory stands an industry that transforms into a filing-and-grant machine: applications are generated by machine, searched by machine, examined by machine, and granted by machine — a procedure that, in routine operation, no longer requires any human interaction. The inventor supplies the core technical information; everything downstream runs automatically.

Litigation, too, becomes a machine matter.

Automation does not stop at grant. In opposition and revocation proceedings, the intensive search for novelty-destroying prior art is the core of every dispute — and this is exactly where AI is strongest. What today ties up weeks of attorney and office work becomes a fast, exhaustive machine search.

A two-tier model becomes conceivable:

  • First instance by AI: The decision on whether a patent is maintained is taken, at first instance, by an AI system — fast, cheap, and based on an exhaustive search.
  • Second instance by humans: Review remains reserved to human examiners — but at significantly higher cost. The human eye shifts from being the standard to being an expensive remedy.

This structure would have far-reaching consequences for access to justice. The cheap, automated first instance lowers the threshold for challenges to IP rights; the expensive human second instance becomes a filter affordable only to those with enough at stake. Whoever wants human review pays for it — a pay-for-human model at the heart of the sovereign procedure.

The Real Fork in the Road

Both scenarios share the same starting diagnosis — the reversal of the cost structure in the office’s favour — and diverge on a single question: Does the EPO give way, or not?

  • If the office stays rigid (Scenario 1), the full force of AI concentrates on the attorney side. Firms consolidate, competition intensifies, and clients benefit only to a limited degree, because the official cost block caps the total saving.
  • If the office gives way (Scenario 2), a largely dehumanised filing-and-grant apparatus emerges, with sharply rising filing numbers — bringing with it all the follow-on problems of quality assurance, a flood of prior art, and the delicate transition to machine-made legal decisions.

Institutional economics argues, in the short term, for Scenario 1. A monopolist without competitive pressure and with its own preservation interests has little incentive to translate efficiency gains into fee reductions. Technical development — particularly the quality of machine search — drives, by contrast, towards Scenario 2.

The likely path lies somewhere in between: the attorney side automates and consolidates first (Scenario 1 is already underway), while the office side follows more slowly, under political pressure, and in stages. The transition to a fully automated procedure would then be no single leap, but a years-long contest over the question of where in the procedure the human remains indispensable — and who pays for them.

What This Means for Applicants and Firms Today

For clients, the decisive cost question shifts. No longer “Which attorney is cheap?”, but “How do I manage the official cost block?” becomes the central strategic task — through the number of claims, the timing of payments, geographic reach, and the question of which rights are even worth maintaining.

For firms, the answer lies not in clinging to the hourly model for text production, which AI is eroding in any case, but in shifting towards what machines cannot (yet) deliver: strategic counsel, the translation of imprecise invention disclosures into robust claims, responsibility and liability for content — and the human second instance that, in Scenario 2, becomes the most valuable asset of all.

The reversal of the cost structure is already reality. Which of the two scenarios follows from it will be decided not in the firms — but on the question of how a monopolistic office deals with the very technology that calls its own cost base into question.

This article is provided for information and discussion and does not constitute legal advice. The fees stated refer to the EPO schedule of fees as adjusted with effect from 1 April 2026.

Posted in European Law, International Law, Patent Law | Tagged , , , | Leave a comment