Anthropic Lets Apple, Amazon Test More Powerful Mythos Model

Anthropic is granting select tech giants—including Apple and Amazon—early access to Mythos, its unreleased, next-generation AI model, ahead of broader deployment. The move aims to assess cybersecurity risks tied to the model’s advanced capabilities, which could enable sophisticated attacks if misused by malicious actors. Testing began in April 2026, with participating firms evaluating Mythos’ vulnerabilities and potential defensive measures before public release. Anthropic’s proactive approach reflects growing industry concerns over AI-driven threats, positioning the company as a leader in responsible AI scaling. This collaboration underscores the urgency of preemptive security testing as AI models grow exponentially more powerful and harder to control.