Sam Altman slams Anthropic's new Mythos model as "fear-based marketing"

OpenAI CEO Sam Altman has publicly criticized Anthropic's new language model, Mythos, calling the company's positioning marketing built on fear. The comment came in Altman's characteristically blunt style and is the latest episode in a drawn-out public standoff between the two largest AI labs.

Anthropic introduced Mythos as a model tuned for cybersecurity tasks. The company pitches it as a tool for defending against threats at a time when AI systems are themselves becoming targets of attacks and are also being used by malicious actors to automate break-ins. The name itself, Mythos, evokes archetypal narratives and dark mythology, which already points to a storytelling strategy with anxious undertones.

Altman sees it differently. In his view, stoking anxiety around AI threats is a tool of competitive struggle, not an honest demonstration of capabilities. Fear sells, as the unwritten rule of industry sales goes, and Altman appears convinced that Anthropic is using it deliberately.

This is not the first time the two camps have traded barbs in public. The rivalry has deep roots. Anthropic was founded in 2021 by a team of OpenAI alumni, including Dario and Daniela Amodei, who left the company over disagreements about safety. Since then, Anthropic has positioned itself as a more responsible alternative, which OpenAI has consistently read as veiled criticism. Mythos fits that narrative: a model built specifically to defend against the dangers of AI is both a product and a manifesto.

The AI cybersecurity market is indeed huge. According to analyst estimates, it will exceed 60 billion dollars by 2030. Large companies, government agencies and financial institutions are looking for ways to defend against attacks that use generative models: next-generation phishing, automated break-ins and deepfakes. With Mythos, Anthropic is aiming squarely at this segment. OpenAI is not standing aside either: the company is developing its own enterprise security tools and actively working with US government agencies. Altman's criticism is not just rhetoric; it is a competitive signal: we do the same thing, just without the apocalyptic narrative.

Public spats between AI industry leaders are becoming part of a new normal. Where tech companies once preferred vague phrases about healthy competition, the heads of the biggest labs now criticize each other openly. Investors, enterprise customers and regulators have to pick sides, or at least form their own position on the thorniest questions of AI safety. Behind the loud statements lies a real technology race, and AI-powered cybersecurity has long since stopped being a hypothetical threat; it is the everyday reality of the corporate world.