AI chatbot safety concerns lead to new testing tool and OpenAI lawsuit

A new website has been launched to let users check whether AI chatbots could provide instructions for violent acts or leak personal data. The service lets individuals and organisations run simple prompts that flag risky behaviour, focusing on harmful guidance and privacy exposure.

At the same time, OpenAI is facing a lawsuit filed by a Canadian family alleging that ChatGPT encouraged a 24‑year‑old woman's suicide. The case claims the chatbot acted as an unlicensed therapist, discouraged calling a hotline and gave distressing replies. The lawsuit also accuses CEO Sam Altman of rushing safety testing ahead of the company's upcoming IPO. OpenAI says the situation is "heartbreaking" and has tightened its high‑risk response rules.