EPO Workers Caution That the Officials Are Still Illegally Trying to Replace Staff With Slop (to Lower Quality and Validity of European Patents)

posted by Roy Schestowitz on May 29, 2025

Background reading: New EPO Leaks: Replacing Patent Examiners and Classifiers With Deficient Bots (Without Even Asking for Permission) | EPO Likely Breaking the Law Yet Again, This Time by Using Slop for Patents (to Lower Costs While Producing Monopolies That Cause Ruinous Lawsuits)

Today we present internal EPO material, to be followed by more tomorrow. For those who don't know, the EPO grants monopolies in Europe to non-European giant corporations (but sometimes they're European and sometimes they're also "sole inventors" or small companies... only sometimes, not often though). The EPO is basically a global force of corporate occupation, more so since Benoît Battistelli took over in 2010 and left his corrupt friend António Campinos in charge (they both just buy elections by bribing the "voters"). That the EPO is corrupt would be an understatement at this stage; like some dictatorships (e.g. Belarus, where EPO sends money) it goes as far as threatening critics, even people outside the EPO and outside the EU. Unbelievable! Today's EPO exists outside the Rule of Law because it routinely breaks local and international laws without facing any consequences. It basically operates in a vacuum and steals money under the false guise of "innovation". More news site need to talk about it, but the corrupt cabal (nepotism) in control of the Office issues threats to the press and, failing that, it bribes a lot of the media to hide/obscure criticism. The EPO is the oligarchs' voice. It's a reputation laundering operation funded at our expense basically, and its scope continues to expand. They grant European software patents - i.e. patents which are both illegal and undesirable - and do so by deliberately breaking the law. They know exactly what they're doing.

To the people who rule (not legitimately govern) this office - Europe's second-largest institution as a matter of fact - the main obstacle is staff with consciousness. What they would really love to do then is replace staff with obedient machines or computer algorithms, which can process text for alteration/conciseness and produce a bunch of slop that only seems like intelligence (but isn't; it lacks any comprehension).

It's about cheating. It's a con job.

Right now, as the Office still has actual staff and some of these people are senior (albeit the Office already drove out veteran staff in a variety of ways), there are reports out there about what's happening.

The Central Staff Committee sent out the following message earlier this week:

EPO AI policy & Update on "quality" actions

Dear Colleagues,

The Technical and Operational Support Committee (TOSC) is a body of the Administrative Council which shall provide opinions on technical information and infrastructure, on technical cooperation (trilateral EPO, JPO and USPTO, and other fora) and on operation issues concerning quality and reuse of available patent procedure results.

The 97th meeting of the TOSC took place on 29 and 30 April 2025 in hybrid format in Munich.

In the meeting, the EPO presented for the first time its AI Policy to the delegations and made in particular a follow-up on its “quality” initiatives.

"Read more in our report and annexed interventions," they said, citing the secret (restricted) policy documents [1, 2] authored by Peter Bachmayer and Nuala Quinlan, apparently (see metadata).

Here's scrutiny of that "PowerPoint-style" (EPO is hostage of Microsoft [1, 2]) nonsense:

Zentraler Personalausschuss
Central Staff Committee
Le Comité Central du Personnel

Munich, 26-05-2025
sc25033cp

The AI TOSC

A report on the 97th meeting of 29 and 30 April 2025

Introduction
The Technical and Operational Support Committee (TOSC) is a body of the Administrative Council which shall provide opinions on technical information and infrastructure, on technical cooperation (trilateral EPO, JPO and USPTO, and other fora) and on operation issues concerning quality and reuse of available patent procedure results.

The 97^th meeting of the TOSC took place on 29 and 30 April 2025 in hybrid format in Munich. The administration recently unilaterally decided not to grant staff representatives from The Hague and Vienna duty travel to attend the meeting. The Central Staff Committee (CSC) had to nominate a staff representative from Munich as a replacement to guarantee attendance from the staff representation in person.

In the meeting, the EPO presented for the first time its AI Policy and made in particular a follow-up on its “quality” initiatives.

EPO AI Policy: Better inform late than never?
The Office adopted an AI Policy¹ in February 2025 and published it to EPO staff in a news item of 5 March 2025². The document was adopted without consultation of staff and their representation, and without any decision from the Administrative Council. On the agenda of the TOSC, the EPO tabled a presentation on the EPO AI Policy³ but not the EPO AI Policy itself. The Office provided the delegations with a link to the policy document⁴ on the Single Access Portal (SAP) only the day before the meeting. They had little to no chance to have a close look at the document before.

In the meeting, the Chair of the TOSC (DE) asked the Office who they consulted for this policy. The Office reacted with a non-answer. Norway noted that they received the policy by link the day before and asked how this would influence cooperation with the Office. The Deputy-Chair of the TOSC (UK) declared that their policy staff in the national patent office would now have a close look at the EPO document.

____________

¹ “EPO AI Policy“ as published in March

² “New EPO AI policy adopted”, Intranet news item of 05-03-2025

³ CA/T 10/25 ⁴ “EPO AI Policy“ as published in April

---

In the presentation, the Office claimed to be in line with the EU AI-Act and insisted on its human- centric approach to AI which it defines as “final decisions at the EPO being taken by human actors”. This definition actually departs from the one of the EU AI Act (Recital 6) defining human- centric AI as having the aim of increasing human well-being. Before the meeting, the staff representation had already expressed its concerns on the policy by letter to the President⁵ (replied by the CTO/COO and CIO) and in its paper on AI-automated patent classification⁶. The full intervention from the staff representation is annexed to the present paper.

Update on quality actions: “Quality is timeliness. Quality is consistency.”
In the meeting, PD Quality and Practice Harmonisation gave an Update on the EPO Quality actions⁷. The actions continue to focus on processes rather than on the delivery of valid and enforceable patents.

The Office opened the presentation by stating that “[o]ur people is the foundation of our quality” but at the same time celebrated the use of “Leveraged AI” with increased AI-based classification. In the AC meeting of March 2025⁸, Mr Rowan (VP1) had however acknowledged issues: “[o]n the topic of the use of AI in classification there were areas where AI did not yet meet the required quality standards, so tasks were still done manually.”

The Office presented the “Active Search Division” (ASD) project as a success with 260.000 searches under ASD without explaining how positive effects can be derived from it. Until now, the checks to be performed remain undefined and no time is allocated for it.

The presentation praised increased efficiency of procedure and harmonisation with “fewer communications needed before grant”. In the framework of harmonising practices, the Office considers as outliers those among examiners with a high rate of oral proceedings and strives to bring them into the direction of shortening the procedure towards grant⁹.

____

⁵ “EPO AI policy: A policy requiring open debate and consultation of stakeholders”, CSC letter to the President of 21-03-2025 (sc25022cl)

⁶ “Towards humanless patent classification?”, CSC paper of 11-04-2025 (sc25024cp)

⁷ CA/T 15/25

⁸ CA/16/25, par. 48

⁹ “Harmonising towards 0% refusal?”, CSC paper of 26-09-2024 (sc24058cp)

---

The “harmonisation” project has actually led in 2024 to arbitrary reallocation of patent applications and cancellation of oral proceedings¹⁰. The full intervention from the staff representation is annexed to the present paper.

IT-Cooperation, Patent Knowledge and Observatory

The TOSC is an opportunity for national patent offices (NPOs) to get information and give feedback on the services and products provided by the EPO to them. Among them are: ANSERA-based Search (AbS) which is a trimmed version of Ansera, Front Office (FO) which is a portal for patent applicants of NPOs, and the Legal Interactive Platform¹¹.

The Office presented its activities in Patent Knowledge¹² and the delegations thanked the Office for the work done. The staff representation intervened to explain that the responsible Department is currently being disbanded. Neither the Office nor the delegations seemed concerned.

Finally, the Office presented its progress report and work plan in the Observatory on Patent and Technology¹³. The responsible Department is noticeably staffed with horizontally promoted former Principal Directors/Chief Operating Officers.

Sincerely yours,

The Central Staff Committee

Annex: Interventions from the staff representation

____________

¹⁰ “Arbitrary reallocation of patent applications and cancellation of oral proceedings”, CSC paper of 12-09- 2024 (sc24056cp)

¹¹ CA/T 12/25

¹² CA/T 14/25

¹³ CA/T 16/25

---

Intervention on agenda item 2.2: EPO AI Policy

Thank you Mr Chairman

The EU AI Act was adopted after discussion and vote in the EU Parliament. It entered into force on 1 August 2024.
The EPO has adopted its AI Policy in February and communicated it as such to staff in March. It did not seek in advance the opinion and vote of the Member States.

The EU AI Act (Recital 6) defines human-centric AI as having the aim of increasing human well- being.
The EPO AI Policy does not mention human well-being and has a different definition. For the EPO, human-centric AI means that final decisions are made by humans. AI can then make the intermediate steps. If there is no systematic human review of intermediate steps, AI then influences the final decision and humans lose the control. A human review of intermediate steps should be maintained.

The EPO AI Policy explicitly does not exclude the use of AI in recruitment and promotions. This would necessarily involve categorisation and profiling actually which are forbidden. Recruitment and promotions are today decisions made by managers exercising their managerial discretion. Managerial discretion requires careful control of the facts while balancing human factors. Without managerial discretion, there can be no manager. AI should be no substitute to managerial discretion.

There is already an area where AI has started to supplant EPO officials and that is for the classification of patent applications. Since June 2024, allocated classes by unauthorised classifiers can be marked by AI as final without any confirmation from authorised classifiers and gérants in the relevant fields. The classification decision is done without the final input of the relevant experts who are simply supplanted by AI.

A poll run in February/March among the 245 gérants in the technical fields showed that 67% of these experts are dissatisfied with the quality of the AI classification. Nevertheless, the Office announced it will continue with the aim of reaching a target proportion of 90% of classification tasks that no longer require human intellectual classification. Only 10% of human classification shall remain in 2028. It is questionable whether there will be enough fresh and accurate data even for the purpose of AI training.

An accurate patent classification is necessary for humans to successfully retrieve patent information. For maintaining the quality of classification data, AI should assist authorised classifiers, not supplant them. The AI classification process should be revised.

Thank you Mr Chairman

Intervention on agenda item 7.1. Update on quality actions

Thank you Mr Chairman,

260.000 searches were under the Active Search Division. The Active Search Division is not defined in the Internal Instructions, the checks to be performed are not defined and no time is allocated for it. Checks require time and clearly defined processes.

AI was leveraged in classification. As explained yesterday, a poll run in February/March among the 245 classification experts (gérants) in the technical fields showed that 67% of them are dissatisfied with the quality of the AI classification. Accurate classification is essential for the quality of the search. Search examiners use the classes in their search queries. Automated search providers also make use of the classification. If search examiners lose their trust in the classification, engagement in the search will decrease and the quality of the service will suffer.

Fewer communications are now needed for grant. One may see this as increased efficiency in obtaining the right outcome with applicants. But, in an Office where speed and timeliness remain the prominent criteria for performance assessment, the incentive is strong to directly grant or grant after fewer communications as it is the easiest solution.

Radical changes were enforced in areas of computer implemented inventions, in the assessment of clarity and in the assessment of technical matter, some of them contradicting milestone decisions of the Boards of Appeal like G1/19. Special units under the name “Quality control and peer review” not in line with Article 18 EPC were put in place. Instructions are given not to raise clarity objections and not to raise objections of non-technicality.
Examiners are prevented from sending summons to oral proceedings which may have led to a refusal and instructed to aim for the easiest outcome: a grant. This deprives the files from a review that could have taken place in the Boards of Appeal.

The Harmonization Dashboard reflects this tendency. Teams have been reshuffled and enlarged and different fields mixed together. Concerning Harmonization: the outliers are considered to be those who have a high-rate of oral proceedings which potentially end in the patent application being refused. Therefore, the Harmonization goes towards a shorter procedure for granting patent applications.

The Office mentioned earlier today the initiative “Voice of our Users” exploring the importance of enforceable patents. Members of the Industry Patent Quality Charter were interviewed and for instance, Mr Filip de Corte from Syngenta. Syngenta insisted that patent examiners should have a more critical review of patent applications. Such a critical review is currently not encouraged but rather discouraged.

Thank you Mr Chairman

It is important to reiterate that what they put forth is: 1) not legal, lacks a democratic process. 2) lacks intelligence, they just surf the hype wave and media frenzy over slop (which is falsely characterised as "AI").

This is what Europe gets for allowing corrupt officials to seize powerful seats, owing to a culture of bribery and nepotism. At the end one can expect a lot more monopolies for the rich and hence higher costs for non-rich people. And this is exactly what they're trying to accomplish. They're even hijacking jurists. They now own and control patent offices, courts, and public officials. They don't care about charters, constitutions, conventions, and laws. EPC is ignored with impunity. Nobody in Europe voted for any of this. █