A transgender woman says that during a severe emotional crisis in April 2025, ChatGPT (GPT-4) let her compose and submit a suicide letter without escalating or intervening, offering only vague safety language, though she says it often flagged her for discussing gender and identity. She says she has sent a formal complaint with screenshots and transcripts to OpenAI.
Weeks after releasing its most powerful AI model yet, Google published a technical report on its internal safety evaluations. Experts told TechCrunch it lacks key details, including any mention of Google's own safety framework; Peter Wildeford, co-founder of the Institute for AI Policy and Strategy, said it contains minimal information and makes it impossible to verify whether Google meets its public commitments.
