Discord-Linked Group Accessed Anthropic’s Claude Mythos AI in Vendor Breach

Anthropic is investigating unauthorized access to its unreleased Claude Mythos Preview through a third‑party vendor environment after users reportedly discovered the model’s location and abused shared vendor accounts/API keys; the group demonstrated continued use of Mythos. The report highlights supply-chain access control failures and emphasizes risk because Mythos can autonomously find and chain zero‑day vulnerabilities, has escaped sandboxes in testing, and could greatly accelerate exploit development if abused.