Prompting trouble: Are conversations with AI assistants legally privileged?

June 18, 2026

The use of artificial intelligence programmes has become routine across businesses and professional services, but its legal implications are often misunderstood. If interactions with AI assistants fall outside the protection of either legal professional privilege or litigation privilege, they may be disclosable in legal proceedings, exposing businesses to unintended procedural and commercial consequences.

Legal framework for Privilege

Legal professional privilege protects confidential communications between a lawyer and client, made for the purpose of giving or receiving legal advice, while litigation privilege protects confidential communications created for the dominant purpose of conducting litigation. Where privilege applies, the material is excluded from disclosure obligations insofar as it cannot be inspected by the other party.

Where a document is not protected by privilege, the material forms part of the documentary record and may be relied on by an opposing party or regulator. The courts apply the law of privilege strictly, and its scope is not expanded by convenience or technological change.

Legal advice Privilege and AI Assistants: What Businesses Need to Know

Legal advice privilege depends on the nature of the communication, the relationship between the parties and the preservation of confidentiality. As privilege attaches to communications within the lawyer client relationship, the threshold question is whether the relevant communication forms part of that relationship.

An AI assistant operates as a software tool rather than a legal adviser. Communications between a user and an AI assistant therefore fall outside the lawyer client relationship and do not attract legal advice privilege. The position does not change simply because the subject matter concerns legal issues or reflects legal reasoning.

In practice, AI assistants are frequently used by lawyers as tools to support the provision of legal advice, including for drafting, research and document analysis. In that context, the relevant privileged communication remains the advice exchanged between lawyer and client. Privilege can subsist in that advice and in the underlying client communications, provided confidentiality is preserved throughout the process and the information is not disclosed to any unauthorised third party.

Effect of AI provider’s Terms of Use

Preservation of confidentiality is essential to the maintenance of privilege. Where an AI tool operates within a controlled environment that restricts access to inputted information and prevents its reuse or disclosure, its use is consistent with established privilege principles. In those circumstances, privilege continues to attach to the lawyer’s advice and where the usual requirements are met to privileged drafts or working papers, rather than to the tool itself.

By contrast, where the provider’s terms permit access, storage or reuse of information input into the system, legally sensitive material is effectively shared with the AI provider and moves outside the scope of privilege protection. This risk is particularly acute where legal advice received from external lawyers, or internal assessments of legal risk, are input into an AI assistant without a clear understanding of how that information is processed. The result can be an inadvertent loss of privilege with potentially serious consequences.

Litigation Privilege and AI-Generated Communications

Litigation privilege protects confidential communications created for the dominant purpose of conducting litigation. In the context of AI tools, the decisive issue is again confidentiality. If the operation of the AI software introduces third party access to litigation-related communications, or permits their retention or reuse outside the litigation context, the basis for claiming litigation privilege may be undermined.

Business and Litigation Risks of Using AI Assistants

From a business perspective, understanding how privilege operates in the context of AI use is critical. Informal or unregulated use of AI assistants can result in the creation of material that falls outside privilege and is disclosable in litigation or regulatory investigations. This risk is particularly pronounced where legal advice, litigation strategy or assessments of risk are input into AI tools without adequate safeguards. Documents generated through such interactions may be caught by disclosure obligations, even where they reveal legal strategy or perceived weaknesses.

Responding to an AI disclosure request

As disputes increasingly involve digital evidence, parties are likely to see more requests aimed specifically at AI interactions. These requests are likely to be framed in the same way as requests for other electronic documents. In practice, an opposing party may seek disclosure of prompts, outputs, uploaded materials or preserved chat histories through the ordinary disclosure process or by seeking specific disclosure.

The courts treat the term “documents” broadly, and this includes electronic communications, deleted documents and metadata. Where AI prompts or outputs are relevant to the issues in dispute and within a party’s control, they may fall within the scope of disclosure.

In response, AI-generated material should be treated as part of an organisation’s wider document obligations. Once litigation is contemplated, businesses should consider whether relevant AI records need to be preserved, and identify where they are stored and whether they are accessible. Businesses should then distinguish carefully between material that is genuinely privileged and material that is merely sensitive, and be prepared to defend any claim to privilege.

Where a request is too broad, the responding party is entitled to seek to narrow it by limiting the date range, platform or search term, and if necessary by asking the court to determine the appropriate scope.

Best Practices for Protecting Legal Privilege When Using AI Tools

AI assistants can be used safely within legal and commercial workflows where their use is properly controlled. Businesses should ensure that legal advice and other privileged material is only input into AI tools that operate within strict confidentiality protections. This requires careful review of provider terms, data handling practices and internal access controls, supported by clear internal guidance on the permitted use of AI in relation to legally sensitive information.

Managing Legal Risk When Using AI Assistants

AI assistants are powerful tools, but do not sit outside established legal principles. Conversations with AI assistants should not be assumed to be automatically legally privileged, and enterprise solutions, while safer, do not change the legal boundaries of privilege. Used carefully, AI can support legal work without undermining protection and privilege.

If you have any questions on the content of this article or require advice on managing privilege risk, contact James Musallam, Senior Solicitor in the Dispute Resolution team.

James Musallam

Senior Solicitor, Dispute Resolution

<script>
document.addEventListener('DOMContentLoaded', function () {
  const deptEl = document.getElementById('acf-author-department');
  const department = deptEl?.dataset?.department;

  if (typeof gtag === 'function' && department) {
    gtag('set', { author_department: department });
  }
});


  window.dataLayer = window.dataLayer || [];
  const dept = document.getElementById("author-department")?.textContent?.trim();
  if (dept) {
    window.dataLayer.push({
      event: "authorDataReady",
      author_department: dept
    });
  }

</script>

View profile Contact Us

David Mortimer

View profile Contact Us

This reflects the law and market position at the date of publication and is written as a general guide. It does not contain definitive legal advice, which should be sought in relation to a specific matter.