AI and patent protection: How public AI tools can put your inventions at risk

As organisations increasingly use generative AI tools to support innovation, product development and innovation-related activities, questions are emerging about the impact of AI on patent protection and confidentiality.

However, what may not be known is that using certain AI tools carries the risk of accidentally revealing confidential know-how or trade secrets, and undermining patent protection. In some cases, this can result in patent applications being rejected, or existing patents being considered invalid from the start and therefore of no value.  

The issue can arise where invention disclosures are entered into, or even produced with, AI systems that are not confidential. This lack of confidentiality essentially makes the invention publicly available. Once treated as public knowledge, the invention may lose its eligibility for patent protection in many regions, including the UK and across Europe. Whilst this remains an evolving area of law, organisations should exercise extreme caution when using AI during research, ideation or the invention capture process.

To help companies and individuals navigate these risks, we've identified the key considerations for using AI in a way that helps preserve confidentiality and, importantly, eligibility for patent protection.

Significant confidentiality concerns #

• Information entered into publicly available AI systems (or tools without confidentiality guarantees) may be retained and reused by the LLM provider, removing control over whether and how long it remains accessible.
• Once confidential knowledge is shared in this way, it risks losing its protected status and may effectively become public.
• This could have significant impact on future patent filings, confidential know-how and/ or trade secrets.
• This is of significant concern if publicly available AI (or AI tools without confidentiality guarantees) are used before filing a patent application, for example, during the invention capture process or even in discussions with patent attorneys about patentability before filing

Human inventorship considerations #

• Patent law requires inventions to be attributed to identifiable individuals.  
• Excessive dependence on AI tools can blur inventorship, making it harder to demonstrate genuine human contribution and weakening enforceability.

Risks linked to AI training data for publicly available AI systems (or AI tools without confidentiality guarantees) #

• Sensitive business information may be absorbed into future AI training cycles, potentially benefiting third parties or competitors.
• Additionally, where AI systems have been trained on already protected works, their outputs may reflect or mirror those materials, creating additional legal and compliance concerns

Mitigating risks when drafting or reviewing information about a potential patent invention #

1. Do NOT enter details of an invention into AI systems before filing a patent application, unless the tool provides clear and reliable confidentiality protections.
2. Make sure you understand and follow your organisation’s rules on the use of AI technologies.
3. Seek advice from a patent professional with experience in both the relevant technology and the legal implications of AI, to assess potential risks.
4. Independently verify the correctness and suitability of any content produced with the assistance of AI.
5. Clearly document all individuals involved in the inventive process and define their respective roles.
6. Keep records that demonstrate genuine human involvement in developing the invention.
7. Engage a patent attorney to ensure the patent documentation is thorough, accurate, and fit for filing.

Organisational risks and safeguards for AI use #

To help organisations realise the benefits of generative AI while minimising legal, regulatory and commercial risks, we recommend the following practical measures:

• Technical controls: Where generative AI is being considered for patent-related activities, restrict use to systems that provide strict data separation and do not reuse your information for model training. Ensure confidentiality protections are in place, and be alert to AI functions that may be switched on by default within everyday software, such as word processors or document management tools.
• Regulatory awareness: Keep track of developing AI-related laws and guidance by following updates from bodies such as the Chartered Institute of Patent Attorneys and the UK Intellectual Property Office, European Patent Office etc. Further information is available in the Chartered Institute of Patent Attorneys' Guidance on the Use of Artificial Intelligence.
• Good AI governance: Put formal policies in place governing AI use, provide appropriate training to staff, record all human input, and retain clear audit records.
• Professional verification: Have all work produced with AI support reviewed by a suitably qualified patent attorney who is familiar with both the underlying technology and the changing legal environment.

As AI technologies continue to evolve, organisations should remain vigilant to the legal, commercial and confidentiality risk they may present. If you have questions about the use of AI in relation to your patent portfolio, invention disclosure processes or wider IP strategy, please contact your usual Murgitroyd attorney or a member of our patent team.