Report: Discord Group Uses Claude's Supposedly Secret Mythos
An unauthorized Discord group accessed Anthropic's supposedly secret Claude Mythos AI model, despite the company reserving it for select partners. The group has not used the model to probe for vulnerabilities, according to a source.
An unauthorized group gained access to Anthropic's Claude Mythos AI model through a private Discord channel. The model was reserved for 'Project Glasswing' partners, including Nvidia, Apple, and Amazon, to find and fix security vulnerabilities. A source told Bloomberg the group used a mix of tactics, including accessing a third-party contractor's account, to gain access. Anthropic is investigating but has found no evidence of unauthorized use beyond the third-party's IT environment. The group has also accessed other unreleased Anthropic models. Anthropic's decision to limit Mythos access was met with criticism from rival OpenAI, which released its own vulnerability-seeking AI model with wider access.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.