G. Law of evidence

In ex parte Fall T 535/21 Im ex parte Fall T 535/21, richtet sich die Beschwerde gegen die Entscheidung der Prüfungsabteilung, die Anmeldung mangels erfinderischer Tätigkeit zurückzuweisen. Die Kammer teilte der Beschwerdeführerin mit der Ladung zur mündlichen Verhandlung ihre vorläufige Meinung mit, dass die Entscheidung zu bestätigen sein würde. In Reaktion darauf legte die Beschwerdeführerin in der mündlichen Verhandlung vor der Beschwerdekammer einen Auszug aus einem Dialog der Beschwerdeführerin mit dem "Microsoft Copilot" vor (Auszug als A1 bezeichnet).

Die Beschwerdeführerin und die Prüfungsabteilung waren sich darüber einig, dass sich Anspruch 1 durch drei Merkmalsgruppen von D3 unterschied. Die Beschwerdeführerin vertrat aber die Ansicht, dass die Unterscheidungsmerkmale mit D3 die synergetische Wirkung hätten, die Gerätesicherheit zu erhöhen.

Was den ersten Unterschied betrifft, hatte die Beschwerdeführerin offengelassen, in welchem Sinne ein Mikrocontroller zur Sicherheit beiträgt.

Zu dieser Frage führte die Beschwerdeführerin aus, dass Mikrocontroller als sichere Steuerung für industrielle Prozesse fungierten, und dass sie echtzeitfähig und aufgrund ihrer geringen Komplexität zuverlässig seien. Die Beschwerdeführerin bezog sich in dieser Hinsicht auf A1. In der dort dargestellten Antwort des Microsoft Copilot auf die Frage "Bewirkt ein Mikrokontroller eine höhere Sicherheit als ein Mikroprozessor oder sind sie gleich bezüglich Sicherheit", wies die Beschwerdeführerin selbst allerdings auf den Warnhinweis in A1 hin, dass "KI-generierte Inhalte [...] falsch sein" könnten.

Die Kammer bemerkte zunächst, dass die in A1 enthaltenen Aussagen per se richtig sein mochten, dass aber A1 kein geeignetes Beweismittel für diese Aussagen war. Zum einen, weil, wie A1 selbst warnte, "KI-generierte Aussagen [...] falsch sein" können. Die Aussagen hätten daher noch durch unabhängige Quellen überprüft werden müssen. Ob sich die in A1 angegebenen Quellen dafür eigneten, falls sie denn überhaupt existierten, hätte ebenfalls geprüft werden müssen. Die Kammer ließ diese Fragen offen, da der Vortrag der Beschwerdeführerin sogar dann nicht überzeugt hätte, wenn der Inhalt der A1 nicht bestritten worden wäre.

Selbst unterstellt, dass Mikrocontroller gegenüber Mikroprozessoren Sicherheitsvorteile hätten, bliebe offen, in genau welchem Sinne die Steuerung des beanspruchten Umrichters durch den verwendeten Mikrocontroller "sicherer" werde. So blieb undefiniert, welches Signal für einen Umrichter sicherheitsrelevant sei oder was im Einzelnen der Benutzer prüfen könnte oder sollte. Die behauptete Synergie war schon deshalb nicht erkennbar.

Das Vorbringen der Beschwerdeführerin reichte nicht aus, um eine Synergie der Unterschiedsmerkmale nachzuweisen. Die Kammer war weiter der Ansicht, dass die Unterschiedsmerkmale für den Fachmann ausgehend von D3 nahe lagen.

In T 449/23, regarding claim 1 of auxiliary request 1 (claims 1 and 2 being identical to claims 2 and 3 of the main request, after claim 1 of the main request was deleted following a finding of lack of inventive step over D5), the board came to the conclusion that the alleged effects of the distinguishing features were not credible, contrary to the arguments of the patent proprietor. Hence, any alleged effects arising from this comparison could not be taken into account in the formulation of the objective technical problem. The patent proprietor also argued that the burden of proof lay with the opponent to demonstrate that the alleged technical effects were not present. The board disagreed, stating:

(a) that the legal burden of proof was the duty of a party to persuade the deciding body of allegations of facts on which the party’s case rested. In principle, a party must prove alleged facts (assertions) from which it infers a legal consequence, i.e. which establish the basis for the party's legal claims. Thus, the allocation of the burden of proof depends on a party’s substantive case.

(b) that to discharge its legal duty of persuasion, a party must prove the alleged facts by appropriate evidence to the required standard of proof. The party with whom the legal burden of proof lies therefore bears the risk that the alleged facts remain unproven, and thus that the deciding body will decide against that party and reject its legal claims. Thus, the legal burden of proof requires the production of appropriate evidence to persuade the deciding body to the required standard.

(c) that in principle the legal burden of proof does not shift. References in the case law to a shift of burden of proof relate to the so-called evidentiary/evidential burden of proof (see for this distinction T 741/91), the notion of which relates to the state of the evidence produced in the course of proceedings. Once the party bearing the legal burden of proof has adduced sufficient evidence to support its allegations of facts to the required standard of proof, the onus is on the adverse party to rebut the asserted facts with appropriate evidence. Otherwise, the adverse party risks that the deciding body is persuaded of the existence of the facts and allows the claims. Thus, if the party having the legal burden of proof has made a "strong case" by filing convincing evidence, the onus of producing counter-evidence shifts to the adverse party. However, this does not mean that the legal burden of proof is on the adverse party to prove the non-existence or the contrary of asserted factual allegations. It is sufficient that the adverse party raises substantiated doubts that prevent the deciding body from being persuaded of the existence of the alleged facts.

(d) that in opposition and opposition-appeal proceedings, each of the parties carries the legal burden of proof for the asserted allegations of facts on which their respective substantive case rests. As regards an alleged lack of inventive step, the burden is on the opponent to adduce appropriate prior art which – when following the established substantive test, i.e. the problem-solution approach – persuades the opposition division or the board of the obviousness of the solution provided by the subject-matter claimed. On the other hand, if the patent proprietor asserts that, in comparison to the prior art, there is an advantage or effect giving rise to a more ambitious formulation of the objective technical problem than that presented by the opponent and hence to an inventive step, the burden of proving this advantage or effect to the required standard of proof is on the patent proprietor. The mere assertion in the patent specification of an advantage or effect cannot be regarded as evidence of such an assertion.

The board listed a number of decisions (T 97/00, T 355/97, T 1097/09, T 1392/04), in which the underlying circumstances were comparable, confirming these principles. The board also observed that T 1797/09 submitted by the patent proprietor remained a singular decision not followed. The subject-matter of claim 1 of auxiliary request 1 lacked inventive step.

In T 733/23 the opposition division had concluded that there had been insufficient evidence to prove that the data sheets D2, D4, and D7 to D9 had been made available to the public before the filing date. Rather than concluding that, as a result of the data sheets not being considered state of the art under Art. 54 EPC, the subject-matter of the claims was novel, the opposition division decided not to admit them into the opposition proceedings. The board concluded that not admitting these data sheets, filed in due time, constituted a substantial procedural violation (see details as from point 4 of the Reasons including discussion on D19, an affidavit).

The board, in support of its decision, presented some key considerations on public availability of advertising brochures and data sheets, as well as the standard of proof to be applied. The board stated that when a document was clearly intended to be publicly distributed, as was the case with advertising or commercial brochures, the absence of a specific publication or distribution date, a situation quite common in this type of document, was not in itself sufficient to conclude that the document did not constitute prior art. As with any other type of evidence, the key question was not whether the exact date of publication could be determined, but whether it could be established that the relevant subject-matter was made available to the public before the priority or filing date.

Data sheets often represent an intermediate case between internal documents and advertising brochures. Where no publication date is present, the board held it should first be assessed whether the document was intended for public distribution. If so, additional sources must be examined to establish whether the relevant subject-matter was publicly accessible before the patent’s filing or priority date. Here, the opposition division had failed to provide a reasoned decision on public availability, giving no weight to the dates printed on their front pages.

As to the standard of proof, the present board concurred with the position in T 1138/20 that there is only one standard of proof: the deciding body must be convinced, based on the underlying circumstances.

According to the present board, this did not imply that all cases were to be treated identically, as in practice the degree of proof required to establish credibility (i.e. to persuade the board) might vary depending on the specific circumstances. In other words, it was not the standard of proof that adjusted with the circumstances, but rather the credibility of the arguments made by the different parties. For example, when the evidence was exclusively controlled by one party, any gaps in the relevant information might significantly undermine that party's credibility. Conversely, when the information was equally accessible to both parties but only one party submitted evidence, merely raising doubts might not be sufficient to challenge the credibility.

In the present case, the conclusions of the opposition division suggested that the standard of proof "up-to-the-hilt" was applied to determine the public availability of the data sheets. Even if the board agreed that different standards should be applied, this would not be justified in the case in hand, as the relevant information to prove the public availability of the data sheets was not within the exclusive sphere of the appellant (opponent). In this instance, the relevant information would more likely be within the sphere of the patentee. Therefore, there was no basis for applying the strict standard of "up-to-the-hilt" or for questioning the credibility of the appellant (opponent) solely on the grounds that some information was missing.

The patentee argued that, when in doubt, the patent should be upheld. The board disagreed. Fact-finding boiled down to a binary exercise: either something had been proven, or it had not. In addition, there was no presumption of patent validity in proceedings meant to re-assess the validity of this very patent.

In T 2463/22 the opposition division had held that the prior uses had not been proven beyond reasonable doubt (up to the hilt), in particular with regard to whether the products of the prior uses had actually been delivered. The parties before the board focused on which standard of proof had to be applied in view of G 2/21 and T 1138/20 and whether the applicable standard had been met. In the respondent-proprietor’s view, T 1138/20 was an isolated decision, not compatible with G 2/21.

On the required standard of proof, the present board observed that G 2/21 recognised that different concepts as to the standard of proof had been developed in the case law. According to T 1138/20 only one standard should be applied, namely "the deciding body must be convinced, taking into account the circumstances of the case and the relevant evidence before it, that the alleged fact occurred".

In the present board's view, under the principle of free evaluation of evidence, it was always decisive in the evaluation of evidence that the members of the deciding body were personally "convinced". Moreover, they had to always be convinced of whether, as stated in T 1138/20, "the alleged fact has occurred". The board stated this was true regardless of which standard of proof was applied. The standard of proof refers to the nature or degree of conviction that the members of the deciding body must have to be satisfied that an alleged fact occurred (see T 832/22).

According to the board, and with reference to a UK House of Lords decision, two important aspects had to be stressed. Firstly, that the standard of proof is related to the required degree of conviction of the members of the deciding body. Secondly, that it is not related to what is evaluated by the deciding body. Hence, also when a lower standard of proof such as the balance of probabilities is applied, the deciding body must assess whether or not the alleged fact indeed occurred. In other words, also when such a standard of proof is applied, the question is not whether the alleged fact might have occurred with some probability. The board considered G 2/21 (points 31 and 45 of the Reasons) consistent with this understanding.

The more specific question as to whether there was only a single standard of proof or more than one could be left unanswered according to the board. The board held that if the deciding body was convinced beyond reasonable doubt that an alleged fact had occurred, there was no need to decide how many standards of proof there were and which one was applicable (see T 832/22).

The board then gave some consideration to the assessment of factual allegations using the beyond reasonable doubt standard of proof. The European Patent Organisation being an independent international organisation, the board stated the standard had an autonomous meaning within this autonomous legal order. Secondly the board agreed with T 832/22 that it seemed expedient to focus on the term "reasonable".

The board then considered the prior uses, focusing especially on prior use relating to the sale of product 5 (sample of a powder mix from a specific lot number), the content of the sample and whether it was available to the public. In view of all the information (including invoices, affidavit, emails, test report, excerpt from database), which also involved evidence provided by a third party (the buyer), the board was convinced beyond reasonable doubt that product 5, with a specific lot number, was sold prior to the effective date of the patent. Since it had also been shown that product 5 disclosed all features of claim 1, lack of novelty prejudiced the maintenance of the patent as granted. Concerning the third auxiliary request, product 5 was suitable for use as closest prior art. The board referred to the reluctance sometimes in the case law to treat an object of a prior use as the closest prior art. Often, there was neither information on what the object did and what properties it had in the technical environment in which it was applied nor on how the process for its manufacture could be modified. These considerations indeed spoke against regarding a prior use as a suitable starting point for assessing inventive step. In the case in hand however, the skilled person was faced with a different situation. The board concluded that the third auxiliary request did not involve an inventive step. The decision of the opposition division was set aside and the patent revoked.

In T 1249/22 the application related to the development – including the training – of an analytical model (e.g. a machine learning model) and the deployment of the trained analytical model on a "compute engine" so as to process live incoming data. The examining division found that the independent claims of the main request lacked an inventive step in view of common general knowledge evidenced by D5.

D5 was a book comprising a collection of individual papers on grid computing, all from different groups of authors, referred to as "chapters" by the editors of the book. The appellant argued that D5 was not evidence of common general knowledge and that each of the chapters of D5 represented a separate piece of prior art; the examining division combined several distinct elements from these chapters without providing any reasoning. The board agreed with the appellant that each of the "chapters" represented a separate piece of prior art, as they appeared to be self-contained papers which did not build on each other, unlike chapters of a textbook. Definitions given in one of these papers did not necessarily apply to the others. D5 rather resembled a conference proceedings volume including a collection of separate papers on a common topic. The mere fact that the papers were published in the same book with a single ISBN did not imply that the whole content of the book formed a single piece of prior art.

As to whether D5, or its individual chapters, were generally suitable as evidence for common general knowledge, the board noted that an allegation that a teaching was common general knowledge might be supported by specific evidence. The deciding body evaluates such evidence by applying the principle of free evaluation of evidence on a case-by-case basis (G 2/21). The board explained that while it might be relevant that the cited evidence was a "book" or a "textbook", this could not, on its own, be decisive, as no firm rules dictate which types of evidence are convincing.

The board further observed that information often appears in a textbook because it was common general knowledge when the book was drafted. However, this did not mean that all textbook content necessarily was common general knowledge or became so upon publication. In the decision under appeal, the examining division referred to Part G, Chapter VII, 3.1 of the Guidelines, in which it was stated that "[i]nformation does not become common general knowledge because it has been published in a particular textbook, reference work, etc.; on the contrary, it appears in books of this kind because it is already common general knowledge (see T 766/91). This means that the information in such a publication must have already become part of common general knowledge some time before the date of publication". The board noted however, that the cited decision T 766/91 only described what is "normally" accepted and what is "usually" the case. In a statement according to Art. 20(2) RPBA, the board explained that the Guidelines had lost this nuance when saying "must" in the passage cited above.

Regarding the examining division's reliance on chapters of D5 as evidence of alleged common general knowledge, the board considered the examining division's reasoning to be insufficient regarding what alleged common general knowledge it was relying on (R. 111(2) EPC). For instance, the examining division referred merely to the "known paradigm of message-based grid computing" without indicating which features of this paradigm were considered to be common general knowledge, despite appearing to rely on more than the knowledge of the existence of that paradigm when considering that all the features relating to the processing pipeline "form part of the common general knowledge of the skilled person".

Thus, the board concluded that the first-instance decision was not sufficiently reasoned within the meaning of R. 111(2) EPC. The case was remitted to the examining division for further prosecution under Art. 111(1), second sentence, EPC and Art. 11 RPBA and the appeal fee was reimbursed under R. 103(1)(a) EPC.