OpenAI CEO Sam Altman stated that competitor Anthropic employed a "fear-based marketing strategy" in promoting its new model, Claude Mythos, and that related security risk warnings may have been exaggerated. Altman pointed out that emphasizing the potential dangers of AI helps justify "a few organizations controlling AI," and while some security concerns do exist, overstating the risks is itself an effective marketing tactic. Previously, Claude Mythos garnered widespread attention for its ability to automatically discover software vulnerabilities and perform complex cyberattack simulations. Anthropic currently offers testing access to a limited number of organizations, including Amazon, Apple, and Microsoft, through a restricted program. Overall, this incident highlights the escalating divergence within the AI industry between a "security-first" and "open development" approach. (Decrypt)