Discord Sleuths Breach Anthropic’s Mythos Preview, Highlighting AI Tool Access Risks
TL;DR
Discord users allegedly gained unauthorized access to Anthropic’s Mythos Preview by analyzing data from a Mercor breach and guessing the model’s online location, aided by permissions tied to an Anthropic contractor; they accessed Mythos and other unreleased models and used the access to build simple websites rather than perform malicious hacks, underscoring ongoing concerns about access controls for AI development tools.
- Security News This Week: Discord Sleuths Gained Unauthorized Access to Anthropic’s Mythos WIRED
- A secretive AI hacking system has sparked a global scramble The Washington Post
- New Security Breaches at Anthropic and OpenAI Proved Mark Zuckerberg Right The Information
- Anthropic’s Mythos Model Is Being Accessed by Unauthorized Users Bloomberg.com
- Can AI Out-Hack Humans? The Dispatch
Want the full story? Read the original reporting
Read on WIRED