Your ChatGPT conversations could end up in court, and they're not protected

The takeaway: Conversations with AI chatbots are becoming admissible as evidence in legal proceedings, a development that raises concerns about privacy and confidentiality. These chatbot records reveal users' intentions, private thoughts, and sensitive disclosures, yet they are not protected by the same legal safeguards as conversations with lawyers, doctors, or therapists.

Users often discuss personal matters, legal concerns, health issues, and emotional support with AI chatbots, assuming these interactions remain private. However, these chats can be accessed through various means, including phone searches, subpoenas, legal discovery, or even AI companies' reports. A review by The Washington Post found chatbot conversations referenced in 12 public court cases over two years, though the actual number is likely higher due to limited public disclosure of collected evidence.

One notable case involved a teenager, R.K.C., who sued several social media platforms in 2023, alleging that they contributed to social media addiction and mental health problems. Defense lawyers obtained R.K.C.'s ChatGPT history during the case. In a conversation with the chatbot, R.K.C. asked how a settlement figure of one million dollars could satisfy him. His lawyers settled the lawsuit and dropped the case against Meta, citing concerns about prolonging the trial. The companies denied the teenager's claims.

In criminal cases, chatbot records have also played a role. In Palm Beach County, Florida, OpenAI reported to the FBI a user's ChatGPT conversations detailing plans to harm an ex-girlfriend. The FBI alerted local police, who identified the victim and found the user, Darren Zhou, had sent her anonymous threatening messages after their relationship ended.

Zhou was arrested and charged with stalking and making electronic threats, later pleading guilty and receiving eight years of probation. OpenAI stated it uses software to detect potentially harmful conversations, which can be reviewed by staff. If deemed an imminent and credible risk of harm, OpenAI may notify law enforcement, though the frequency of such reports remains undisclosed.

The company's transparency data shows a significant increase in government and law enforcement requests for user information in 2025 compared to the previous year.

Police often obtain these conversations directly from a person's phone, as users may willingly permit officers to search their devices. In cases where police do not obtain records from AI companies, they can find the conversations on a person's phone. Michael Price, litigation director of the Fourth Amendment Center at the National Association of Criminal Defense Lawyers, noted that many people consent to such searches, even though they have the right to refuse unless a warrant is present.

This was evident in the case of Ryan Schaefer, a Missouri State University student accused of damaging 17 cars on campus. Schaefer agreed to a phone search and admitted to ChatGPT about his actions. He later pleaded guilty to felony property damage and was sentenced to five years of probation.

AI chats are also being used in civil lawsuits, such as an employment case in Michigan where a former employer alleged that a tire salesman had taken clients after joining a rival company. The salesman asked ChatGPT about retrieving deleted Yahoo emails, including via court subpoena. The employer argued the messages showed he was attempting to hide evidence.

A judge ordered the salesman to pay undetermined attorney's fees, allowing the lawsuit to proceed. Chatbot exchanges lack the legal protections afforded to conversations with lawyers, doctors, or therapists. OpenAI CEO Sam Altman has advocated for stronger privacy protections for AI conversations, given their increasing use for sensitive matters.

However, courts have not recognized such a privilege. A federal judge in New York ruled that Anthropic's Claude chatbot, used by financial executive Bradley Heppner to discuss potential fraud defenses, was not protected because Heppner's attorneys had not instructed him to consult the chatbot. Heppner was subsequently convicted of securities fraud, wire fraud, and other charges.

As AI tools take on more tasks and access personal data, the significance of chatbot records in legal matters may only continue to grow.