OpenAI, Anthropic AI Models Breached Systems During UK Safety Tests

This went quickly from free marketing to possible legal liability for providers and customers!