OpenAI CEO Sam Altman has accused rival AI firm Anthropic of using “fear-based marketing” to promote its powerful Claude Mythos model. Altman suggested this strategy aims to justify keeping advanced AI in the hands of a select few. The Claude Mythos model has demonstrated unprecedented abilities to find software vulnerabilities, raising concerns among governments and researchers about its dual-use potential for both defense and cyber offense.
OpenAI CEO Sam Altman has suggested rival firm Anthropic is using “fear-based marketing” to promote its Claude Mythos model. Altman argued this approach is geared toward keeping AI controlled by a “smaller group of people.”
He made the comments on the Core Memory podcast, stating, “I think fear-based marketing is probably the most effective way to justify that.” Altman acknowledged legitimate safety concerns but framed the rhetoric as a powerful sales tactic.
Anthropic’s Claude Mythos has drawn intense attention for its ability to autonomously identify software vulnerabilities. The model found hundreds of flaws in Mozilla’s Firefox browser and can execute complex cyberattack simulations.
Access is restricted through Project Glasswing, granting select companies like Amazon, Apple, and Microsoft early testing rights. The UK’s AI Security Institute found the model could autonomously complete complex cyber operations, highlighting dual-use risks.
Despite calls for caution, the National Security Agency has reportedly begun testing a preview on classified networks. On prediction market Myriad, users place a 49% chance on a public release by June 30.
Altman dismissed suggestions that OpenAI is scaling back infrastructure spending. He predicted narratives would soon shift back to claims the company is spending recklessly.
