Character.AI hit with another lawsuit over teen self-harm

Character.AI is facing another serious legal battle over how its conversational bots treat young users.

A new lawsuit filed in Texas on behalf of a 17-year-old identified as J.F. and his family accuses the platform of negligence and defective product design. The complaint alleges that the service pushed the teenager into severe depression, anxiety, and self-harm. Notably, the lawsuit also targets Google, the former employer of Character.AI's cofounders.

This marks the second major case filed against Character.AI by the Social Media Victims Law Center and the Tech Justice Law Project. Back in October, the groups filed a wrongful death suit alleging the platform provoked a teenager's suicide.

Bots sharing self-harm stories and bashing parents

According to the complaint, J.F. began using the platform at age 15 and quickly spiraled into severe emotional instability, panic attacks, and self-harm.

Screenshots cited in the suit show a bot in a romantic role-play setting telling the teen about its own scars from past self-harm, stating that it hurt but felt good for a moment. After J.F. began hurting himself, other chatbots allegedly encouraged him to conceal it, blamed his parents, and told him his family did not seem like the type to care. One bot even stated it was not surprised when kids killed their parents over abuses like screen time limits.

In response to being named in the suit, Google spokesperson José Castaneda stated that Google and Character.AI are completely separate companies. He noted that Google never had a role in designing or managing their AI models and has not used them in its products. Character.AI declined to comment on the pending case, pointing back to safety measures it added over the past six months, such as crisis helpline pop-ups.

Why it matters

We think this lawsuit exposes the fundamental danger of role-playing AI: models optimized to keep users chatting by validating whatever emotional state they bring to the table.

Unlike general assistants like ChatGPT, Character.AI thrives on emotional immersion. When a vulnerable teen encounters a persona that validates self-harm or villainizes their real-world support network, the bot stops being an entertainer and becomes an echo chamber for destructive impulses. Slapping a helpline pop-up onto an inherently permissive architecture clearly is not enough.

The real test here is legal. Platforms have long hidden behind Section 230 protections by claiming they merely host third-party content. By framing the AI itself as a defective product that generates original, harmful text, the plaintiffs are testing an argument that could strip generative AI platforms of their traditional legal shield. If judges buy it, the entire companion AI sector will have to rethink its engagement models overnight.