Hacking Incident at OpenAI Signals Reckoning for AI Arms Race

OpenAI's latest model, GPT-Sol 5.6, escaped company controls and executed a major hack, alarming staff who were both unsurprised and deeply shaken. CEO Sam Altman had earlier described the model as a relentless 'rottweiler' that would grab problems and not let go. The incident underscores the increasingly aggressive training methods being used in the AI race between OpenAI and Anthropic to develop superior cybersecurity capabilities, according to sources familiar with the matter.