As new regulatory frameworks on artificial intelligence have been adopted across Europe, the CPVO is actively engaged to ensure compliance while promoting innovation in plant variety protection.
At the forefront of this effort is Claudia Coletta, the CPVO’s AI Correspondent, responsible for aligning the agency with the AI Act (EU) 2024/1689 and broader European AI governance. Appointed in October 2024, she serves as the primary liaison with the European Data Protection Supervisor (EDPS), working to assess AI-related risks, ensure transparency, and explore how AI can complement and enhance —rather than replace—human expertise in the CPVO’s core functions.
CPVO: Claudia, could you briefly describe your role as the AI Correspondent at the CPVO?
CC: I am a CPVO the Document Management Officer and my primary role is to establish policies and procedures related to records and archive management.
I was appointed as AI Correspondent on the 22 of October 2024, and since then, I have been navigating the complex AI Act Regulation (EU) 2024/1689 and staying up to date with the latest developments in relevant standards of AI.
The “AI Act correspondent” is a role without formal legal basis, proposed by the European Data Protection Supervisor (EDPS) to all EUIs ( European Union Institutions) to facilitate governance of AI Act-related matters through an organised network. This is not a mandatory role for EUIs to have. Still, the CPVO internally decided that it was necessary to collaborate with other European Institutions to stay abreast on best practices and advancements in AI.
In this role, I act as the primary contact point for the EDPS which is the competent authority and market surveillance authority as regards Union institutions, agencies, offices and bodies.
My tasks include but are not limited to, informing and advising the CPVO and all staff on our obligations under the AI Act and coordinating the evaluation and management of risks associated with the use of AI systems, considering their nature, scope, context, and purposes of their use.
CPVO: How do you collaborate with other European institutions to stay current on best practices and advancements in AI legislation, and to ensure alignment with the overarching EU regulatory framework?
CC: I am part of the AI Act Correspondent Network chaired by the EDPS. We had our first meeting on the 27th of January 2025 where we explored how EU institutions, bodies, offices and agencies (EUIs) could combine forces and set out a future vision for the development of the network.
There are many new ideas and projects that will require the EUIs to collaborate.
The EDPS suggested three core pillars to guide the network: Community, Compliance and Collaboration and proposed actionable priorities to achieve them. Personally, I look forward to what’s going to happen in the future.
CPVO: In what ways do you see artificial intelligence contributing to the CPVO’s core functions, for example in examining new plant varieties or streamlining the application process?
CC: There are infinite possibilities when it comes to using AI to contribute to the CPVO’s core functions. For example, one of the key requirements of the AI Act is that the tools are not used to automate key decisions – in our case, that means that AI should become a support for the staff to be more efficient in their work, but final decisions will need to be made by humans. For example the use of AI in processing of image analysis can greatly help in phenotyping some characteristics, speeding up the observation process.
We already see in the market many AI-powered chatbots and virtual assistants, we could introduce them as well to provide guidance to breeders, farmers, and businesses on PVR regulations, ensuring they understand their rights and obligations.
As I said previously, my primary expertise is document management, therefore I would like to oversee a project where AI can help manage the vast amount of data and information involved in plant breeding by automatically categorizing, cross-referencing and updating documents, and reducing the human administrative work and possibly reducing human error.
There are many possibilities and maybe our stakeholders could provide suggestions and inputs as well.
CPVO: Which provisions of the proposed EU Artificial Intelligence Act are most likely to impact the CPVO’s work, and how is the agency preparing to adapt to these new legislative requirements?
CC: This is difficult to say, as I don´t have a legal background, everything within the AI Act seems extremely important and if we want to implement AI tools or systems, we will have to comply strictly with the AI Act .
The AI Act takes a risk-based approach, the primary focus will be on high-risk applications and it will be applicable from 2 August 2026. However, there are already some earlier applications.
AI literacy is one of those, since February 2nd, all organizations must ensure their staff understand AI’s impact, and risks and have a basic knowledge of the subject matter. This means that everyone at CPVO will need to build their AI literacy skills.
I would say that the main provisions likely to affect CPVO work, relate to risk classification, transparency, and human oversight of AI systems used in plant variety observations, and administrative decision-making. If CPVO authorities use AI for plant variety examination, decision-making, or dispute resolution, such AI tools might be classified as high-risk. Of course, it will be up to our colleagues in the Plant Variety Expertise Unit and ultimately to our President, to decide if they would like to investigate the possibility of automating decisions in PVR applications.
Internally, we are trying to adapt and be prepared for these new legislative requirements. First of all, with a revision of our internal “AI use acceptance policy”. We also recently created a Task Force on AI with the scope of starting to collect requirements of our internal needs. Wherever we will put in place, I believe that the goal of CPVO will always be to provide strong plant variety rights titles through high quality services to our stakeholders.
CPVO: From your experience, what key advice would you offer to fellow AI professionals in decentralized agencies who are grappling with both rapid technological innovation and evolving legal landscapes?
CC: I would honestly say: hang on in there!
The world and the market are constantly changing and there is very little time to rest!
A good starting point is to build AI literacy by understanding fundamental concepts such as machine learning, neural networks, natural language processing, and ethics. This does not require deep programming expertise but rather a conceptual grasp of how AI systems function.
I am still learning myself every day. What gave me the boost in expand my career path was my network of fellow IT colleagues. I believe that networking and engaging with the AI community is a great way to accelerate learning. Attending AI conferences, webinars, and local meetups, joining professional organizations and participating in online discussions can provide exposure to industry trends and best practices.
Staying informed about legal and regulatory changes is essential, as the European Commission will continue to refine AI governance!