Unknown Group Allegedly Accesses Claude Mythos, Anthropic Investigates Security Hole
TL;DR
Anthropic says an unknown group accessed Claude Mythos—an unreleased, highly restricted model—via a third‑party vendor environment. Bloomberg reports, citing a live demo and screenshots, that the group used data from a Mercor breach and other intel to pinpoint Mythos and has been experimenting with it since April 7. The group claims no malicious intent, but Anthropic is investigating the breach and the security hole remains a concern.
Topics:businessartificial-intelligence#anthropic#artificial-intelligence#claude-mythos#cybersecurity#data-breach#unauthorized-access
- Some Unknown Group Is Reportedly Using Claude Mythos Without Permission Gizmodo
- Anthropic’s Mythos Model Is Being Accessed by Unauthorized Users Bloomberg.com
- Anthropic investigating unauthorised access of powerful Mythos AI model Financial Times
- Faster attacks, quicker exploits: Early testers of new AI models warn of their new abilities Axios
- Sam Altman throws shade at Anthropic’s cyber model, Mythos: ‘fear-based marketing’ TechCrunch
Want the full story? Read the original reporting
Read on Gizmodo