AI tools such as ChatGPT and Claude are increasingly used in day-to-day business. While they can be helpful, using public AI tools in connection with legal work creates significant risks. This applies not only to disputes, but also to transactional and advisory work, particularly where a dispute may arise later.

Courts have warned that using public AI tools may result in loss of confidentiality and waiver of legal professional privilege. This means that documents or communications that were intended to be private may have to be disclosed to opponents or the court.

The risks arise with public, open-access AI tools only. They do not apply to enterprise AI systems that have been contractually vetted and approved for confidential use.

  • What is legal professional privilege?

    Legal professional privilege protects confidential communications between a client and their lawyers for the purpose of giving or receiving legal advice or conducting litigation. If confidentiality is lost, privilege is lost with it.

    Clients are responsible for preserving privilege in their own communications. This includes how they handle legal advice and documents, whether in a dispute or in the course of general commercial work.

    Even in non-contentious matters (such as M&A, real estate transactions or commercial contracts), documents and advice may later become relevant in a dispute. If those materials have been shared with a public AI tool, privilege may already have been lost before the dispute arises.

  • What are the courts saying?

    Judicial guidance in this area is developing rapidly, but the direction of travel is clear.

    In a recent decision, the Judge said: “Uploading confidential documents 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”.

    In a US case, the Court reached a similar conclusion and, whilst not binding in the UK, it highlights the issues arising and the wider approach being taken. In this case, a client used a public AI chatbot (Claude) to analyse advice he had received from his lawyers, explore his legal exposure and defence strategy, and generate documents relating to his case. He later shared those AI generated documents with his lawyers. The Court held that:

    • inputting privileged advice into public AI tools can waive privilege
    • sharing non-privileged AI-generated documents with your lawyer does not make those documents privileged
    • using an AI tool does not amount to legal advice – these public tools expressly disclaim that purpose and recommend consulting a lawyer
  • What about AI notetakers and transcription tools?

    The same risks apply to AI-powered meeting tools, including notetakers, transcription services and meeting assistants. Many of these tools process and store meeting data in ways that may not preserve confidentiality.

    If privilege is to be maintained, care must be taken before allowing any such tool to access meetings where legal advice is discussed or sensitive documents are reviewed.

  • What should you do (and not do)?

    • Do not upload legal advice, communications, draft documents, transaction materials or dispute-related information into public AI tools. Even where the information is anonymised, the risks remain.
    • Do not use AI notetakers, transcription services or meeting assistants in legal discussions without checking how those tools handle data.
    • Always speak to your legal advisers before using AI tools in connection with legal matters.

Key takeaways

  • Treat anything inputted into a public AI tool as published to the world.
  • Privilege depends on confidentiality – once lost, it cannot be recovered.
  • Check with your legal adviser first.

Get in touch

For more information on this topic or any relating, contact our experts below or meet our team here.

Meet the team Read more about Get in touch