AI has become firmly embedded in the research and development processes of many companies, enabling teams to automate certain tasks and increase their technical productivity. These benefits are real. Yet the question of AI and intellectual property in R&D is often overlooked. What happens to the inventions arising from these processes? Can they be protected, defended and valorised — and under what conditions?
This is the central question that Adrien Marcone, European and Swiss Patent Attorney and specialist in intellectual property strategy for the digital world (AI/ML, blockchain, cryptography, IoT) at P&TS, addressed in his presentation on 1 September 2026, at an event organised by CoBooster in collaboration with Sylvac and AI Swiss.
An Event Dedicated to the Industrial Practice of AI in R&D
Held at Sylvac‘s premises in Yverdon-les-Bains, this half-day event brought together more than 40 participants around seven presentations delivered by eight specialists, complemented by two workshops and a guided tour of Sylvac’s facilities. The event was opened by Eric Schnyder, CEO of Sylvac, Joëlle Tosetti, Managing Director of CoBooster, and Stéphane Fallet, President of AI Swiss.
CoBooster is a collaborative platform supporting the creation of project teams, the co-funding of feasibility studies and the transfer of patents within the Swiss industrial ecosystem. P&TS is one of its partners.
Alongside Adrien Marcone, the following speakers took part:
- Cristina Apetrei and Cosmin Cioroiu (Philip Morris International), on how AI is reshaping discovery, development and decision-making;
- Nicolas Souvlakis (Embiggen X), on AI solutions for technology sourcing and procurement;
- Bertil Chapuis (HEIG-VD), on the human, organisational and practical dimensions of AI adoption;
- Houssem Ben Salem (Sylvac), on the integration of AI agents in industrial environments;
- Julian Nolan (Iprova), on AI-assisted invention in the age of human-machine collaboration;
- Vincent Nassar (IntoFlow), on leading responsible AI adoption in R&D.
Four Intellectual Property Considerations to Anticipate When Using AI in R&D
In his presentation, Adrien Marcone identified four issues that any company risks facing when deploying AI in its R&D processes without having first defined a solid intellectual property framework. These situations are relevant to both technical teams and legal departments, regardless of the size or sector of the organisation.
1. Your R&D Secrets May Already Be Exposed
Submitting R&D data to third-party AI tools directly exposes a company’s trade secrets and confidential information. The risk takes on a particular dimension from a patent perspective: even partial disclosure of an invention prior to filing a protection application can definitively compromise any prospect of patent protection.
To manage this risk, it is essential to:
- Not assume that an AI tool processes data confidentially by default, and to systematically verify the data policies of each tool used;
- Seek formal contractual agreements with providers, beyond standard terms and conditions;
- Define an internal AI usage policy and train teams accordingly;
- Keep a record of all tools used, along with their respective data policies;
- Explicitly address the use of AI tools in confidentiality agreements and collaboration contracts.
2. AI Created It. But Who Owns It?
Under the law, AI cannot be designated as inventor or author. It is the human contribution — documented and traceable — that determines any possibility of claiming ownership.
To secure ownership of inventions, we recommend:
- Documenting human contributions at every stage of the R&D process;
- Bearing in mind that AI-generated code benefits from weak or no copyright protection;
- Ensuring the traceability of human input throughout the development process;
- In collaborative R&D contexts, explicitly addressing the treatment of AI-generated outputs in contractual IP clauses.
3. Your AI Outputs May Infringe Existing Rights
The outputs of an AI system tend to approximate the state of the art, which increases the risk of inadvertently infringing existing patents. There is also a methodological difficulty: AI is not always a reliable tool for assessing the state of the art in patent matters.
To manage this risk:
- Conduct a Freedom to Operate (FTO) analysis before commercialising any AI-assisted invention;
- Involve an intellectual property adviser from the earliest stages of the R&D process;
- Document the verification and human refinement steps applied to AI outputs.
4. Free Code, Costly Consequences
AI code-generation tools have been trained on codebases covering a wide range of licences, including copyleft licences (GPL, LGPL…). Code subject to such obligations may inadvertently become incorporated into a proprietary codebase. In the worst-case scenario, a company may be legally required to release its entire codebase as open source.
To prevent this, it is important to:
- Favour AI tools that are transparent about their training data, in particular the licences under which that training data is distributed;
- Systematically document AI-generated code within the development process;
- Establish clear development guidelines for teams;
- Maintain or introduce the use of code-scanning tools (FOSSA, Black Duck, etc.);
- Retain human oversight at every stage of the process.
Three Guiding Principles for Protecting What AI Contributes to Creating
In summarising his presentation, Adrien Marcone put forward three principles that capture the issues developed throughout his talk:
- AI tools do not guarantee any intellectual property protection by default.
- Ownership of an invention rests on documented and traceable human contribution.
- Integrating an AI tool into an R&D process is also a legal decision.
These three principles are not solely relevant to large companies with dedicated legal departments. They apply to any organisation that innovates with AI, regardless of size or sector, and that wishes to be in a position to protect, defend and valorise what it creates.
It is precisely to address these challenges that P&TS has developed, within a team of complementary technical and legal profiles, solid expertise in the field of IT and AI law, serving the companies and academic institutions that innovate in the digital world.
Photos: @CoBooster & P&TS
