OpenAI fixes flaw after Claude helps researchers reach internal code

OpenAI fixed a flaw after Hacktron researchers used Anthropic's Claude to compromise employee accounts and reach an internal GitHub repository within 72 hours.