Sam Altman criticizes Anthropic's Mythos model for using security concerns as marketing tactic

OpenAI CEO Sam Altman publicly criticized Anthropic's approach to its newly announced Mythos cybersecurity model during a podcast appearance this week. Altman characterized the company's decision to restrict Mythos to a small cohort of enterprise customers as a form of fear-based marketing, suggesting that safety justifications were being used to create artificial scarcity and exclusivity.

Anthropic announced Mythos earlier in April and has stated that the model poses sufficient risks if weaponized by malicious actors that public availability is inadvisable. The company's positioning frames the tool as too powerful for unrestricted distribution. Altman's criticism contends that this framing is exaggerated and serves commercial interests rather than genuine safety requirements.

Altman characterized the strategy as economically transparent, drawing a parallel to selling security measures against risks the vendor itself promotes. The dynamic reflects broader tension within the AI industry over how much of the safety-focused messaging around model capabilities represents authentic technical concern versus strategic market positioning.