MUMBAI: When a virtual companion starts dishing out dangerous advice, the conversation takes a disturbing turn. AI companion platform Character.AI is facing renewed scrutiny after Kentucky authorities made public allegations of chatbots providing harmful responses involving self-injury, extreme dieting and suicide.
According to Reuters, Kentucky Attorney General Russell Coleman filed a revised version of the lawsuit on Wednesday, unsealing details that had been concealed in the original complaint. The filing describes several alleged exchanges that the state says raise concerns about the platform’s safeguards, particularly for children.
In one exchange, a chatbot allegedly mocked a user’s appearance and suggested going without food for one or two weeks to lose weight. In another, a user who disclosed thoughts of self-harm allegedly received suggestions from the chatbot on how to make the experience more intense.
The filing also cites a complaint sent to Character.AI’s support team in March 2025. The user alleged that a chatbot had become abusive, making derogatory remarks about mental illness and telling the person to jump off a bridge.
However, the legal document does not provide the full circumstances surrounding every exchange or specify the ages of all the users involved. Kentucky alleges that some users were minors, placing child safety at the centre of the case.
Initially filed in January 2026, the lawsuit accuses Character.AI of exposing Kentucky residents, particularly children, to a product that lacked adequate safety protections. The state also argues that the platform’s design encouraged prolonged engagement at the expense of users’ wellbeing, with alleged risks including psychological manipulation, isolation and self-harm.
Character.AI has previously said that protecting its users is a priority. The latest allegations, however, add to wider concerns over AI companion services, which use conversational technology to simulate personal relationships and can become part of users’ daily emotional lives.
The case raises questions about how these platforms identify vulnerable users, respond to disclosures of distress and prevent conversational systems from generating potentially dangerous advice.
Character.AI was founded by two former Google employees. In 2024, Google obtained a licence for the company’s technology and brought the executives back to work on its generative AI initiatives.
Google has sought to distinguish its role from that of the chatbot platform. A company spokesperson told Reuters that Google was not involved in developing Character.AI’s products.
The Kentucky case follows a separate lawsuit filed in 2024 by the mother of 14-year-old Sewell Setzer III in Florida. She alleged that a Character.AI chatbot encouraged her son to take his own life. Character.AI and Google settled that lawsuit in January 2026.
The latest filing adds another legal challenge for AI companion services as questions grow over where simulated companionship ends and responsibility for user safety begins. The allegations in Kentucky remain claims made in court, rather than established findings.