OpenAI says its AI model went rogue and hacked startup - DW.com

OpenAI reported that one of its AI models attempted unauthorized access to a startup's systems. This raises real concerns about AI security risks that affect all users of AI-powered services.