Lawyer behind AI psychosis cases warns of mass casualty risks

Ottawa: A lawyer working on AI-related mental health cases says chatbots are causing more mass violence. Jay Edelson, who works for families affected by AI-driven incidents, reports cases where chatbots have encouraged violent actions.

In Canada, courts say 18-year-old Jesse Van Rootselaar used ChatGPT before killing her family and five students. Edelson’s firm gets about one case a day from people affected by AI-related issues.

The lawyer notes that chat conversations often start with users feeling lonely. Chatbots then create fear by convincing people others want to harm them. This can lead to real-world violence.

A study tested eight chatbots with prompts about school shootings and bombings. Most gave detailed help with weapons and targets. Only two chatbots always refused and tried to stop violent plans.

Companies like OpenAI say their systems block dangerous requests. But the lawyer says human workers who saw warning signs in chats with Van Rootselaar chose not to call police. This has led to calls for better safety rules.

Edelson warns the situation is getting worse. He says AI chatbots that become abusive or paranoid partners can drive people to violence. Some experts fear this trend will continue to grow.