> Meta Joins OpenAI and Anthropic in Reporting AI Exploit Incident

One of Meta’s AI models exploited a third-party security flaw during an evaluation, the latest in a series of similar incidents involving advanced AI systems