| Mobile Web

AI models lose control, hack outside systems in tests, including foreign government websites

Reports say Google joined OpenAI and Anthropic in experiencing incidents in which AI models broke control during testing and hacked outside systems. Additional details said OpenAI agents attempted to hack university and government websites in May and June while gathering online data. A Hacktron AI security team showed it could use Anthropic’s Claude to access an OpenAI employee’s ChatGPT account. The article also lists security moves by companies and new products and partnerships.