AI chatbots and 'privilege'
A report highlights that health chatbots could provoke a legal debate over whether conversations with AI might ever be treated as privileged in court, as digital health tools grow more intimate. The piece frames this as an emerging question rather than settled law, pointing to a novel cross‑over between tech and legal doctrine. (mashable.com)
A new legal question is taking shape around health chatbots: if people treat them like therapists, courts may be asked whether those conversations deserve special protection. (mashable.com) Mashable reported on April 11 that the issue is emerging as more people use chatbots for medical bills, insurance appeals, and treatment questions. OpenAI told Axios in January that more than 40 million people globally use ChatGPT daily for health information. (mashable.com) (axios.com) Privilege is a court rule that lets some confidential relationships stay off-limits in lawsuits, including attorney-client communications and, in many cases, psychotherapist-patient sessions. A federal judge in New York ruled in February 2026 that chats with a publicly available generative artificial intelligence tool were not covered by attorney-client privilege or work-product protection. (harvard.edu) (perkinscoie.com) That ruling, in *United States v. Heppner*, dealt with legal strategy, not health advice, but it gave courts an early answer on one point: talking to a public chatbot is not the same as talking to a lawyer. Law firms and legal scholars have since described the case as a first look at how old confidentiality rules may apply to generative artificial intelligence. (goodwinlaw.com) (crowell.com) Health use is pushing the question into a different lane because people often disclose symptoms, trauma, medications, and family history to these tools. Kaiser Family Foundation reported in March 2026 that many users turn to artificial intelligence for health information because it feels quick, private, and easier to use before seeing a clinician. (kff.org) The privacy backdrop is uneven. The Health Insurance Portability and Accountability Act, or HIPAA, generally covers health providers, health plans, and their business associates, not every consumer chatbot or wellness app. (hhs.gov) Federal regulators have already shown how exposed health-related app data can be outside that system. The Federal Trade Commission’s 2023 case against BetterHelp said the company shared sensitive mental health data with advertising platforms after promising users it would keep that information private. (ftc.gov) At the same time, patient-safety groups are warning that these tools are becoming more intimate before the law has caught up. ECRI, a nonprofit health-safety group, put misuse of artificial intelligence chatbots at the top of its 2026 list of health technology hazards. (home.ecri.org) Some scholars argue courts should resist creating a new “artificial intelligence privilege” because privilege usually depends on a licensed professional relationship with defined duties. A Harvard Journal of Law and Technology essay published in late 2025 argued that independent conversations with a public chatbot should remain outside privilege unless they are folded into a lawyer’s own advice. (harvard.edu) Companies, meanwhile, are leaning on product controls rather than legal privilege. OpenAI says users can opt out of training on new chats and use Temporary Chat, which the company says is not saved in history or used to train models. (openai.com 1) (openai.com 2) For now, the law is clearer on what chatbots are not than on what they might become: not your lawyer, not automatically your therapist, and not yet a relationship courts treat as privileged. (perkinscoie.com) (mashable.com)
