Altman criticizes Anthropic's restricted Mythos cybersecurity model as 'fear-based marketing'

OpenAI CEO Sam Altman called out competitor Anthropic's approach to its new cybersecurity model Mythos, describing the company's restricted release strategy as "fear-based marketing" during an April 21 appearance on the "Core Memory" podcast.

Anthropic released Mythos earlier in April 2026 to a limited cohort of enterprise customers only. The company claims the model is too powerful to release publicly, citing concerns that cybercriminals could weaponize it. Specific capabilities, pricing, and technical specifications have not been disclosed.

Altman's criticism

"There are people in the world who, for a long time, have wanted to keep AI in the hands of a smaller group of people," Altman said. "You can justify that in a lot of different ways."

He added: "It is clearly incredible marketing to say, 'We have built a bomb, we are about to drop it on your head. We will sell you a bomb shelter for $100 million.'"

Altman implied that Anthropic's safety justifications serve to maintain AI access for an exclusive elite while generating marketing impact.

Industry context

Critics have questioned whether Anthropic's concerns about Mythos are proportionate to the model's actual capabilities. No independent verification of the model's cybersecurity capabilities has been published.

The AI industry broadly has used apocalyptic rhetoric to market products. Both AI safety advocates and company executives—including Altman himself—have made public statements warning about AI's potential to cause catastrophic harm, while simultaneously developing and selling the technology.

The OpenAI-Anthropic rivalry has intensified in recent months, with both companies competing for enterprise customers and positioning themselves as leaders in AI safety and capability.

What this means

This public dispute highlights the tension between AI safety claims and commercial interests. When companies restrict model access citing safety concerns, it becomes difficult to verify whether those concerns are technically justified or serve as competitive positioning. Anthropic's enterprise-only release strategy for Mythos could be genuine risk mitigation or, as Altman suggests, a way to command premium pricing from customers who believe they're accessing uniquely powerful technology. Without technical details or independent evaluation, the cybersecurity capabilities of Mythos remain unverified claims.