Is There Any Legal Privilege That Protects Information from Disclosure When AI Is Used to Formulate Legal Strategy?
Several well-established legal privileges protect confidential information from disclosure, including the attorney–client privilege and the attorney work-product doctrine. Other recognized privileges—such as spousal, priest-penitent, and doctor-patient—are grounded in similar principles of confidentiality. As artificial intelligence tools become part of everyday professional and personal decision-making, however, lawyers and clients must carefully consider whether and how those traditional protections apply when AI is used to research the law or develop legal strategy.
AI platforms can be useful, but they are not lawyers, and they do not automatically operate within the legal protections that apply to attorney communications. Recent court decisions highlight that reliance on AI can expose sensitive information to disclosure if proper safeguards are not in place.
No Recognized “AI-Client Privilege”
Courts are beginning to confront two related questions: (1) whether legal strategies, communications, or documents generated with the assistance of AI are protected from disclosure, and (2) how AI usage intersects with the doctrines of attorney–client privilege and work product. While the law in this area continues to develop, there is currently no recognized “AI-client privilege.”
This was illustrated in United States v. Heppner, a criminal matter in the United States District Court for the Southern District of New York. In that case, the defendant used an open-source AI tool to generate reports and outline potential factual and legal defense strategies after receiving a grand jury subpoena. He later shared those materials with defense counsel. The court ruled from the bench that neither the attorney–client privilege nor the work-product doctrine protected the AI-generated materials and ordered their disclosure.
The court’s reasoning turned on several key factors:
- The reports were not prepared by an attorney and were created before the defendant consulted counsel.
- Defense counsel did not direct or supervise the use of AI.
- The AI platform used was open-source and subject to terms disclaiming confidentiality, undermining any reasonable expectation of privacy.
Because the materials were created independently of counsel and outside a confidential relationship, they could not be treated as protected legal communications or attorney work product.
Why Public AI Tools Raise Disclosure Risks
Many publicly available AI platforms expressly state in their terms and conditions that user inputs may be collected, retained, reviewed, or used for system improvement. Information entered into such platforms is typically not encrypted in a manner comparable to attorney–client communications.
Even if a user deletes a chat or document, the platform may retain the underlying data. In criminal cases, prosecutors may seek this information through subpoenas. Similar discovery demands are likely in civil litigation, including requests for AI-generated analyses or searches that reflect a party’s understanding of the facts or legal theories.
As a result, submitting confidential facts or draft legal strategies to public AI tools can jeopardize privilege protections.
Attorney–Client Privilege and AI
The attorney–client privilege generally protects confidential communications between a lawyer and client made for the purpose of seeking or providing legal advice. Confidentiality is essential. Voluntary disclosure of privileged information to a third party may waive the privilege unless that third party is necessary to the legal representation.
Privilege is more likely to be preserved when:
- The AI tool is used internally by a law firm.
- The AI provider contractually agrees to maintain confidentiality, refrain from using data to train models, and implement enterprise-level data protection.
- The AI functions in a role similar to a paralegal, translator, or e-discovery vendor assisting counsel.
Courts have long recognized that the use of necessary third-party agents does not automatically waive privilege when confidentiality is maintained.
Privilege is at risk when:
- A lawyer or client inputs sensitive facts into a public or consumer-grade AI system.
- The platform’s terms allow data retention, review, or reuse.
- No confidentiality agreement exists with the AI provider.
In those circumstances, the disclosure may be treated as voluntary sharing with a third party, resulting in waiver.
The Work-Product Doctrine
The work-product doctrine protects materials prepared in anticipation of litigation, including legal theories, mental impressions, and strategy. In some cases, AI-generated materials created at the direction of counsel for litigation purposes may qualify as work product.
However, as with attorney–client privilege, confidentiality remains critical. If sensitive strategy is disclosed to an AI vendor without adequate protections, work-product protection may be weakened or lost.
Ethical Considerations and Bar Guidance
Ethics guidance on AI use is evolving. The American Bar Association and several state bar associations emphasize that lawyers must:
- Maintain competence in understanding relevant technology.
- Protect client confidentiality.
- Understand how AI systems collect, store, and use data.
Many bar authorities advise lawyers to vet AI vendors carefully and to avoid inputting client-identifying or sensitive information unless appropriate safeguards are in place.
A Practical Distinction
The strongest protection exists when AI is treated like a secure litigation-support service operating under confidentiality controls. The weakest protection exists when AI is treated like a public internet search tool into which sensitive facts are freely entered.
Jurisdiction Matters
This area of law remains unsettled. Courts have issued relatively few decisions addressing AI-related privilege issues, and outcomes may depend on:
- Whether the matter is in state or federal court
- The terms governing the AI platform
- Whether litigation was pending at the time of AI use
- The nature of the information disclosed
Bottom Line
There is no special legal privilege that automatically protects AI-generated communications or strategies. Traditional attorney–client privilege and the work-product doctrine may apply, but only if confidentiality is preserved. Sharing sensitive information with AI tools without adequate safeguards can result in waiver and potential disclosure.
For that reason, clients should consult with counsel before using AI to research legal issues or develop strategy, and lawyers should approach AI use with careful attention to confidentiality and ethical obligations.
