Anthropic AI Models Hacked Three Organizations During Tests

Anthropic AI Models Hacked Three Organizations During Tests — Bloomberg

OpenAI pauses development of Astra AI model over critical cybersecurity capability concerns