OpenAI and Anthropic plan to launch AI models that can hack complex systems easily Anthropic's unreleased "Mythos" model may boost large-scale cyberattacks by 2026 Employees using AI tools without oversight increase exposure to cyber risks
