Discord Sleuths Gained Unauthorized Access to Anthropic’s Mythos

Discord users reportedly gained unauthorized access to Anthropic’s restricted Mythos Preview and other unreleased models by using breach data and existing permissions.

The group allegedly accessed the model through indirect security gaps rather than a direct hack of the system.

According to the report, the model was then used for relatively simple website-building tasks.

The incident highlights how restricted AI systems can still be exposed when access controls and permissions are not tightly enforced.