FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Patent and trade mark attorneys
Putting unfiled invention details into an AI tool that is not fully secure may make them available to the public and destroy novelty. For attorneys, the same habit can also put statutory privilege and client confidentiality at risk.
Part of: AI in patent and trade mark attorneys
In short
Under the Patents Act 1977 s.2, an invention is new only if it is not part of the state of the art, which includes anything made available to the public in any way before the priority date. CIPA's December 2025 guidance warns that entering invention details into an AI tool that is not fully secure may amount to public disclosure. Attorney communications have statutory privilege under CDPA 1988 s.280 and TMA 1994 s.87, and confidentiality underpins privilege. Use tools with contractual confidentiality, no training on your data and controlled retention. A leak may engage both cyber and PI cover.
Under section 2 of the Patents Act 1977, an invention is new only if it does not form part of the state of the art. The state of the art includes all matter made available to the public "by written or oral description, by use or in any other way" before the priority date. A pre-filing disclosure, even an accidental one, can therefore be fatal.
CIPA's AI guidance for SMEs and creators (4 December 2025) applies this to AI directly: entering invention details into an AI tool that is not fully secure may amount to public disclosure and lose patentability in the UK and Europe. The guidance is aimed at clients, which means clients may ask their attorney which tools are safe, and may expect the attorney's own tools to meet the same standard.
Communications with patent attorneys have statutory privilege under section 280 of the Copyright, Designs and Patents Act 1988, and trade mark attorney communications under section 87 of the Trade Marks Act 1994. Both put attorney communications on the same footing as a solicitor's.
Confidentiality is a prerequisite for privilege. In UK and R (Munir) v SSHD [2026] UKUT 81 (IAC) at ¶21, the Upper Tribunal said uploading client documents to a public AI chatbot put them in the public domain, breaching confidentiality and waiving privilege. The Chancellor of the High Court's speech of 22 April 2026 said public AI tools appear not to be confidential, while with secure tools it is "hard to see" any impact on privilege. The speech framed questions rather than deciding them, and neither source concerned IP attorneys specifically, but the reasoning is directly relevant.
Illustrative scenario: an inventor sends a detailed disclosure to a firm before filing. To save time, a trainee pastes it into a free generative AI chatbot to produce a first draft of the background and summary. The chatbot's terms allow inputs to be retained and used to improve the service. Months later, the client's competitor opposes the granted patent and, during disclosure, the firm's AI use comes to light. Whether the upload was a public disclosure would be argued on the facts, but the client now faces cost and uncertainty, and blames the firm.
Similar risks arise when draft claims or confidential infringement analysis are pasted into AI translation or summarising tools.
A leak can create two kinds of loss. The firm's own costs, such as investigation and notification, usually fall to a cyber policy. A client's claim that the firm negligently destroyed novelty or waived privilege is a civil liability claim, which usually falls to PI. IPReg's Minimum Terms and Conditions include express affirmative cover for claims arising from a Cyber Act or Cyber Incident (cl.1.1.7), with related exclusions at cl.6.5 and 6.6; how a given wording applies to an AI upload depends on its definitions. See cyber and PI overlap for AI content and cyber exclusions in PI.
Related guides: AI patent drafting, AI prior art searching, does attorneys' PI cover AI mistakes and the attorneys hub.
How Apex helps: Apex arranges PI for patent and trade mark attorneys, and a broker can check how both policies would respond to an AI data leak.
It may. The state of the art includes matter made available to the public in any way before the priority date (Patents Act 1977 s.2). CIPA warns that entering invention details into a tool that is not fully secure may amount to public disclosure. The answer will depend on the tool's terms and facts.
Privilege depends on confidentiality. The Upper Tribunal in Munir said uploading client documents to a public AI chatbot waived privilege. Secure tools with contractual confidentiality are much less likely to raise the issue, but the courts have not settled every question.
Often both. The firm's own response costs usually fall under cyber cover, while a client's negligence claim usually falls under PI. IPReg's minimum terms include affirmative cyber wording at cl.1.1.7, but the outcome depends on each policy's definitions and exclusions.
Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start a PI quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.
