[DigitalToday reporter Yoonseo Lee] Sam Altman (샘 알트먼), OpenAI's chief executive officer (CEO), has directly criticised Anthropic's new cybersecurity model, "Claude Mythos" (Claude Mythos).

On April 21 local time, IT outlet TechCrunch reported that Altman appeared on the podcast "Core Memory" and claimed Anthropic is stoking fear to make the product appear more threatening than its actual performance.

The core of his remarks centred on how Anthropic released the model earlier this month. Anthropic has made the model available only on a limited basis to some corporate customers. The company has said a public release is difficult due to concerns that cybercriminals could weaponise it. Criticism has also been raised that such explanations are exaggerated.

Altman described the approach itself as "fear marketing". He said there have long been people who wanted to leave artificial intelligence (AI) within a small, exclusive group, and that such an approach can be justified in various ways. He argued that a strategy of narrowing access while emphasising a model's risks could ultimately concentrate control of the technology in the hands of a few.

His rhetoric escalated further. Comparing Anthropic's messaging, Altman said it was "great marketing" to say, "We built a bomb and it will soon drop on your head. Instead, we'll sell you a $100 million air-raid shelter." He criticised it as an approach that highlights risks to increase the scarcity and perceived need for its product.

The dispute shows competition between OpenAI and Anthropic spreading beyond product performance to a battle over messaging on safety and the scope of public release. Anthropic has cited the potential for misuse as the basis for not making Claude Mythos available to general consumers, and Altman countered that the logic is closer to a means of persuading the market.

Some point out that such an approach is not unfamiliar across the AI industry. Warnings that AI could become overly powerful or increase societal risks have been repeated not only by critics of the technology but also by companies that sell it. It is also mentioned that Altman is not completely free of that trend.

In this situation, the remarks are seen as an example that goes beyond checking a rival and again shows how generative AI companies explain safety in connection with business strategy. That is because in fields such as cybersecurity, where concerns about misuse are high, limiting release, selecting customers and stressing risks can become part of a product strategy. The war of nerves between OpenAI and Anthropic appears likely to continue, intertwined with debate over AI safety and access control.