Conversations with artificial intelligence chatbots are increasingly being sought and used as evidence in legal proceedings, raising concerns about user privacy and the perceived confidentiality of these interactions. Unlike communications with legal or medical professionals, which are typically privileged, exchanges with AI systems like OpenAI's ChatGPT and Anthropic's Claude lack such protections and can be compelled for disclosure.
Over the past two years, at least a dozen court cases have cited chatbot logs. Data from OpenAI indicates a significant increase in the disclosure of user account content in response to legal requests. The company reported disclosing content from over 80 user accounts in the latter half of 2025, a more than fourfold increase compared to the second half of 2024.
These disclosures often occur when users consent to searches, such as a university student who, after damaging 17 cars, handed over his phone to police, revealing his panicked queries to ChatGPT about potential detection. In another instance, the ChatGPT history of a teenager suing tech companies over social media addiction was drawn into discovery.
Even attempts to delete chat histories may not guarantee privacy. In a copyright lawsuit filed by The New York Times against OpenAI, a judge ordered the AI company to preserve chat logs, including those users had requested to erase. OpenAI stated that this requirement forced it to forgo privacy protections it had implemented, despite its agreements to delete data under the EU’s General Data Protection Regulation (GDPR) and California’s privacy laws.
Beyond legal requests, AI companies may proactively share user data with law enforcement. OpenAI's policy permits its reviewers to refer conversations to authorities if they identify an "imminent and credible risk of harm to others." One such incident involved OpenAI contacting police in Palm Beach County, Florida, after a ChatGPT user repeatedly described plans to harm an ex-girlfriend.
This trend extends beyond AI chats, mirroring a broader pattern of technology companies providing user data to law enforcement and litigants. Between 2014 and 2024, Google, Meta, and Apple collectively shared details from 3.16 million U.S. user accounts, with annual increases in the number of records disclosed. Previous cases have involved subpoenas for audio recordings from smart speakers, such as an Amazon Echo in a 2019 murder investigation, and private messages from social media platforms, as seen in a 2022 illegal abortion case where Facebook provided messages between a mother and daughter to police.
Users are advised to treat all chatbot conversations as potentially discoverable records, even if they feel informal or if a delete function is available. It is recommended to avoid sharing sensitive information with AI that one would typically discuss only with a doctor or lawyer, given the lack of privilege and confidentiality protections for AI interactions. This caution also applies to other digital communications, including social media posts and private messages, which can also become evidence in legal contexts.
