“Discussing an invention with an AI tool where the terms of use do not include an obligation to keep information confidential carries a very significant risk of being considered a disclosure to the public”
AI can help develop ideas and explain complex technology. It can also help with understanding the law. But before sharing confidential information with an AI, such as an unpublished invention, it is imperative to check whether the information will remain confidential. The law is developing, and the consequences of getting this wrong could be significant.
A recent UK judgment of the Upper Tribunal of the Immigration and Asylum Chamber found that uploading confidential client documents to a public AI tool placed them in the public domain, breaching confidentiality and waiving legal privilege.
The judgment stated, at §21, that: “to put client letters and decision letters from the Home Office into an open source AI tool, such as ChatGPT, is to place this information on the internet in the public domain, and thus to breach client confidentiality and waive legal privilege, and thus any regulated legal professional or firm that does so would, in addition to needing to bring this to the attention of their regulator, be advised to consult with the Information Commissioner’s Office. Closed source AI tools which do not place information in the public domain, such as Microsoft Copilot, are available for tasks such as summarising without these risks”.
Open-source versus closed-source AI tools
This case does not bind the High Court (which has jurisdiction over patents in England and Wales) but is likely to be considered a persuasive line of legal argument. Notably, the judgment draws a distinction between ‘open-source’ and ‘closed-source’ AI tools, which is clearly a misnomer that seems to mean ‘non-confidential’ and ‘confidential’ respectively.
Discussing an invention with an AI tool where the terms of use do not include an obligation to keep information confidential carries a very significant risk of being considered a disclosure to the public, which would invalidate a later filed patent in the UK. In practice ‘enterprise’ AI tends to include these confidentiality obligations, and ‘consumer’ tiers of AI are less likely to.
International context
In the United States there has been a couple of legal cases where this has been tested. In United States v. Heppner, a defendant’s independently prepared AI exchanges attracted neither attorney-client privilege nor work product protection. In Warner v. Gilbarco, Inc., however, a self-represented litigant’s AI-assisted materials retained work product protection. These were not patent-novelty decisions, but they caution against blanket assumptions about legal protection.
Meanwhile in Europe the European Patent Office (EPO) does not have any cases from the boards of appeal on this subject at the time or writing.
Check before you upload
These are the four steps you should take before sharing confidential technical information on AI:
- Establish what confidentiality obligations bind the provider
- Can inputs be used for training or service improvement
- Who can access them, and how long are they retained
- Does the proposed use comply with obligations to clients and collaborators
Remember that a paid subscription or “not used for training” promise does not answer every confidentiality question. The applicable terms, settings and technical safeguards need careful review.
If you have already shared an invention with an AI, preserve the details and seek advice promptly. Do not assume either that patent rights have been lost or that everything is safe.
The law is not settled. That uncertainty is a reason to check confidentiality carefully before using AI, not a reason to take the risk.
Contact your usual Barker Brettell patent attorney for advice before sharing unpublished technical information.
