Opinion from PCC on Interpretation of the Guidelines on AI
The Professional Conduct Committee (PCC) provides opinions upon enquiries from epi members under Art. 7(c) of the epi Code of Conduct. Any opinion given does not have regulatory force and is prepared with the intention to provide helpful assistance. No liability of any kind attaches to the epi, the Professional Conduct Committee or any members of that Committee in respect of these opinions. In accordance with Article 7(c) CoC, opinions of the Professional Conduct Committee shall not be binding on the disciplinary bodies. The following opinion has been considered useful for epi members as the questions it addresses are particularly significant. Hence, it has been decided to publish it, in anonymised form.
Summary of Enquiry
“According to the „epi Guidelines: Use of Generative AI in the Work of Patent Attorneys”, members must establish, in advance of using generative AI, the wishes of their clients with regard to the use of generative AI.
Question 1: Can this requirement be fulfilled by including a clause in the general terms of engagement? For example: “The member and their staff may use AI tools unless the client explicitly objects in writing for a specific case. The use of AI tools does not diminish the member’s responsibility regarding diligence.”
Would this approach place the burden on the client? If this is a desirable approach, how should it be phrased to our clients?
Question 2: Would the situation be improved if a firm adopted an internal AI code of conduct, which would be available to clients upon request? Additionally, do we have examples of AI codes of conduct that we recommend to our epi members?“
Relevant Provisions
- Regulation on Discipline for Professional Representatives, Article 1
- epi Code of Conduct, Articles 1(c), 2(a), 3(a), 3(b), 3(d) & 3(e)
- epi Guidelines: Use of Generative AI in the Work of Patent Attorneys, Guidelines 4, 5a & 5b
Opinion
Question 1
Referring firstly to Article 1 of the Regulation on Discipline, this clearly articulates that a professional representative shall exercise his professionArticle 1 further says that when exercising his profession a professional representative must proceed conscientiously, but this seems to be simply a condition pertaining to how the representative undertakes the tasks in question. The requirement for conscientiousness in the view of the Chamber does not detract from the responsibility of the representative to undertake or oversee the work requested by a client.. This in the opinion of the Chamber means a professional representative must take responsibility for ensuring both that the outcome of their work is in a significant way the result of the representative’s efforts, and that the process of carrying out the work is under the control of the representative.
It follows from this that a representative cannot delegate control of the work steps to an unqualified third party, whether in human or machine form.
In the opinion of the Chamber “control” does not necessarily refer to carrying out all of the steps by the representative. Hence a representative legitimately can satisfy the responsibility requirement even if their work amounts to checking of work produced by others. However such checking should in the opinion of the Chamber include knowing at least the key features of the process by which the work product is prepared; and it should never be merely a trivial aspect of the work process.
The obligation of the professional representative to retain responsibility for work processes is reinforced by the wording of the Code of Conduct. Article 1(c) of this requires a Member (i.e. a professional representative) to serve as a reliable adviser and act as an independent counsellor. These conditions cannot be met if a machine undertakes the totality of the work. At the least therefore the Member must inform themselves about the processes used by the machine, and must check that the resulting work is of the same standard as if a human had completed it.
The requirement for the representative to exercise their profession does not mean that the representative is forbidden to use tools such as AI products in the course of their work. The use of such tools is clearly contemplated by the epi Guidelines, and as a generally stated principle the Chamber believes that using AI tools is not inconsistent with the requirements of Article 1 of the Regulation on Discipline.
Guideline 4 however recommends that generative AI tools should only be used when this is in line with the wishes of a client. Guideline 4 further recommends a representative to establish those wishes in advance of using an AI tool.
Guideline 4 in the view of the Chamber means that permission to use (for example) AI tools should not be given in a generalised way.
On the contrary, giving general permission to use AI tools means that the client may not truly know the extent to which such tools are being used by a representative or for which aspects of the representative’s work. Any permission given by a client in this circumstance would not meet the requirement to establish the wishes of the client in advance of using an AI tool.
Moreover the granting of generalised permission runs the risk of failing to take account of technological developments in the AI tools. As an example, at the time a generalised authority is given to use an AI tool it may not include a claim-drafting module. The introduction of such a module at a later date may (without changing the identity of the AI tool) fundamentally change how a representative uses the tool. A generalised authorisation is unlikely to take account of this.
Overall the granting permission to use AI tools through operation of a default feature of terms of engagement of a private practice firm, without stating the areas in which those tools are to be used, does not appear to the Chamber to satisfy the requirement for the representative to establish the wishes of a client in advance of using the AI tools.
However the use of a default agreement is not inherently unacceptable as long as the conditions it creates are clear and explicit. In other words if the terms and conditions forming the core of a default agreement to use AI tools identify the areas in which the tools are to be used this would seem to provide adequate safeguards for clients in the areas identified above.
If the terms are made available via a website this has the additional advantage that any change in the uses to which the tools are to be put can readily be communicated.
As a voluntary observation the Chamber notes that Article 1 of the Regulation on Discipline requires a professional representative not knowingly to make any false or misleading statement, a concept that is reinforced in Articles 2 and 3 of the Code of Conduct. Private practice firms can help to avoid any accusation of such behaviour if their terms of engagement when discussing AI tools make clear the extent to which such tools are to be used.
Articles 3(d) and 3(e) of the Code of Conduct set out the obligations of epi Members with respect to the activities of non-Member colleagues. These sections of the Code were drafted before artificial intelligence was a recognised area of technology. The Chamber however believes it would be illogical not to apply the principles of Articles 3(d) and 3(e) (mutatis mutandis) to work carried out using AI tools; and indeed the Guidelines (Annex 3) were drafted with Articles 3(d) and 3(e) in mind.
In summary the Chamber is of the opinion that a generalised agreement over the use of AI tools is not sufficient to satisfy the requirement of Guideline 4 unless the terms identify the areas of work in which AI tools are to be used. This conclusion seems to be relevant whether the agreement is achieved by way of a firm’s terms of engagement stated on a website, or in another way.
Question 2
The chamber considers that implementing an internal AI code of conduct and stating in a public way that this exists is favorable and can be recommended.
The Chamber wonders whether restriction of access to detailed information on the use of AI tools only to clients was inadvertent. The result of this is that the only information available to potential clients amounts to a generalised agreement set down in the Terms. However as explained above this does not appear to be adequate to allow potential clients to make informed decisions, and in the view of the Chamber requires augmenting to indicate the areas of work in which AI tools will be used.
The fact that non-clients seemingly do not have access to comprehensive information, with an internal code of conduct only being made available once a party has agreed to become a client, in the view of the Chamber moreover increases the risk that the published terms may include or amount to a misleading statement.
In summary therefore an “internal” AI code of conduct, that is not made available until after an entity becomes a client of a firm, does not assist in causing a set of website terms to comply with the principles set out above in response to Question 1. As indicated, at least the main features of such a code should be available to non-clients, in order to allow them to make informed decisions over whether a firm’s AI policies are acceptable.
The presentation of clear information on the likely use by private practice firms of AI tools is recommended, and it is strongly recommended that such information provides those wishing to engage the firms with well-founded confidence that such use will comply at least with Article 1 of the Regulation on Discipline and preferably also Guideline 4 explained above.
The Chamber does not presently have examples of internal policy documents, pertaining to the use of AI products, that could be made available for use by Members of epi. However the Chamber suggests that one aspect could relate to the levels of skill, experience and training of staff members responsible for the use of AI tools in private practice firms. Hence a useful provision might be one indicating that those responsible for the selection and use of AI tools must be sufficiently familiar with both the nature of the AI tools and the work of a European Patent Attorney as to be confident that such use will comply with the epi Guidelines forming Annex 3.
Summary of Opinion
In summary, the Chamber believes that compliance with the epi Guidelines on use of AI can be achieved through the inclusion of consent provisions in terms of engagement in a firm’s website. However such terms in the opinion of the Chamber should identify the principal areas in which generative AI is to be used.
Moreover, more detailed information on the likely areas of use should be readily available in the event of an enquiry to the firm. Such detailed information should not be restricted only to entities that are clients of the firm in question, and should be available so that clients and potential clients alike can make informed decisions about whether to accept the firm’s AI usage policies.
The Chamber does not presently have examples of internal policy documents, pertaining to the use of AI products. However one aspect could relate to the levels of skill, experience and training of staff members responsible for the use of AI tools in private practice firms, particularly with regard to ensuring compliance with the epi Guidelines on the Use of Generative AI in the Work of Patent Attorneys.
This opinion does not have regulatory force and is prepared with the intention to provide helpful assistance. An opinion provided in accordance with Article 7(c) CoC is not binding on the Disciplinary Bodies. No liability of any kind attaches to epi, its Professional Conduct Committee or any members of that Committee in respect of this opinion.
