An increasing number of people are treating artificial intelligence (AI) chatbots as “confidants” to whom they can confide in about emotional issues, health troubles, work secrets, and more. However, real-life cases have shown that these seemingly private chat conversations are gradually making their way into courtrooms and becoming evidence in civil and criminal cases.
A recent investigation by The Washington Post in the United States, based on public court records and local news reports, revealed that over the past two years, at least 12 civil or criminal cases have cited dialogue records from AI chatbots.
According to court documents, a teenager using the pseudonym R.K.C. sought explanations from the chatbot ChatGPT because he did not understand his father’s words.
“My dad said I would get $1 million in compensation,” R.K.C. said to ChatGPT in October 2024. “He said, if this doesn’t make me happy, then what will? What does he mean by that?”
This question from R.K.C., along with other private conversations he had with ChatGPT, was later made public and included in the court records by the defense lawyer.
In 2023, R.K.C. filed lawsuits against Meta, Snapchat, TikTok, and YouTube, alleging that the apps from these companies had led to his social media addiction over the years and had caused mental health issues.
In late July this year, R.K.C.’s appointed lawyer agreed to settle with Snap, TikTok, and YouTube and dropped the lawsuit against Meta. The lawyer stated that the decision to discontinue the litigation was due to concerns about the teenager’s mental and emotional well-being enduring “weeks of grueling trial.”
One of the law firms representing R.K.C.—Morgan & Morgan, one of the partners, Mike Morgan—said in a statement, “A 15-year-old may type things into a chatbot that he would not say to a therapist, parent, or friend, and he does not realize that this content may eventually end up in the reports of expert witnesses for the defense.”
R.K.C. is just one among many Americans who are now finding their private conversations with AI chatbots potentially becoming public in courtrooms.
Americans pose millions of questions to AI chatbots daily. As people become more accustomed to revealing personal thoughts and secrets to AI, these platforms have accumulated a vast amount of sensitive data that could be accessed by law enforcement agencies, courts, or litigants.
In some cases, AI companies may even proactively report users on the platform to the Federal Bureau of Investigation (FBI), leading to investigations and potential prosecutions of the individuals.
In May of this year, the FBI notified the Palm Beach County police in Florida that OpenAI, the developer of ChatGPT, had reported a man identified in court documents as Darren Zhou, who had been confiding in ChatGPT about his plans to harm his ex-girlfriend.
Police confirmed the identity of the woman based on the information provided by the FBI and found that Zhou had been anonymously sending disturbing messages to her since their breakup, including messages like “can’t wait to smell your scent.”
Zhou was arrested in May on charges of stalking and electronic threats. Police stated in court documents that the messages he had sent to his ex-girlfriend and to ChatGPT confirmed the threats and were not empty threats.
He pleaded guilty this month and was sentenced to eight years of probation. Zhou’s lawyer declined to comment on the case.
OpenAI had previously stated that the company would use a system to detect conversations that might involve dangerous behaviors and have them reviewed by humans. If the reviewers find credible and urgent risks of harm to others, the company may report them to law enforcement agencies.
In another case, a university student from Missouri was suspected of vandalizing 17 cars in the campus parking lot. When the police investigated him, they obtained his phone, in which he had asked ChatGPT on the night of the incident if he was “in big trouble” and inquired if the police could trace back to him if he had damaged multiple cars.
These chat conversations were later deemed crucial evidence by the police. The man eventually admitted to a felony property damage charge and was given a five-year probation.
Another case involved a tire salesman in Missouri in a civil lawsuit against his former employer. The former employer accused him of violating a non-compete agreement after resigning and trying to claim the previous company’s clients.
During the evidence disclosure process, lawyers conducted a comprehensive check on his electronic devices and found that he had asked ChatGPT whether emails deleted years ago could still be retrieved by Yahoo and if the related emails could be obtained even if a court subpoena was received.
Based on this discovery, the employer questioned whether he had hidden or destroyed evidence. The judge ultimately ruled that these behaviors constituted strong evidence suggesting he may have intentionally withheld information and ordered him to pay part of the legal fees incurred from additional investigation.
Currently, conversations with AI chatbots like ChatGPT typically do not enjoy the legal privilege of confidentiality between a lawyer and a client.
As reported by The Washington Post, a financial industry executive under federal investigation for alleged financial fraud had sought potential defense strategies from the AI model Claude. He argued that these discussions should be protected similarly to attorney-client communications and thus beyond the prosecution’s scrutiny.
However, a federal judge in New York ruled in February that this assertion was invalid because AI is not a lawyer, and the individual’s actual attorney did not request such discussions with the chatbot.
In the wake of suicides by some ChatGPT users and shooting incidents involving ChatGPT users in Canada and Florida, OpenAI stated that they would be more proactive in reporting potentially concerning conversations to authorities.
Legal experts point out that chatbots create an atmosphere that encourages users to candidly reveal personal privacy, and therefore, the records held by AI companies may be richer than traditional digital evidence.
“This includes not just the questions asked by the user but all the prompts, background information, and the entire conversation flow. Your thought process and intent are laid bare, with no need for speculation,” said Michael Price, the Litigation Director of the Fourth Amendment Center under the National Association of Criminal Defense Lawyers (NACDL).
He added, “If there is a window into the soul and life privacy, this seems to be it.”
Data released by OpenAI indicates a rapid increase in requests from government agencies and law enforcement departments for user data. In the latter half of 2025, the company disclosed data from over 80 accounts, more than four times the number compared to the previous year.
Andrew Ferguson, a law professor at George Washington University and author of “The Rise of Big Data Policing: Surveillance, Race, and the Future of Law Enforcement,” warns that as AI agents gain more personal privacy permissions and integrate further into public life, more aspects of people’s lives will be exposed to digital surveillance.
He reiterates similar concerns in his latest book, “Your Data Will Be Used Against You: Policing in the Age of Self-Surveillance,” stating that modern individuals are engaging in a form of “self-surveillance.”
“Your entire world is now open to law enforcement,” he said.