curl Creator Dismisses Anthropic's Mythos AI as Overhyped: 'Primarily Marketing'

Daniel Stenberg, the creator and lead maintainer of the ubiquitous curl software, has published a blistering assessment of Anthropic's Mythos AI model, declaring that the tool's much-hyped code analysis capabilities are mostly promotional noise. In an exclusive blog post, Stenberg states bluntly that Mythos does not deliver a breakthrough in finding security vulnerabilities beyond what other modern AI tools already achieve.

The Core Verdict: Hype Over Substance

"My personal conclusion can however not end up with anything else than that the big hype around this model so far was primarily marketing," Stenberg wrote. He evaluated Mythos against the curl source code repository and found no evidence that it identifies flaws at a "higher or more advanced degree" than other AI-powered analyzers. "Maybe this model is a little bit better, but even if it is, it is not better to a degree that seems to make a significant dent in code analyzing."

Stenberg's assessment directly challenges Anthropic's earlier claims that Mythos was too dangerous for public release—a narrative that fueled widespread media interest. He argues that the AI's performance on curl's codebase was unremarkable, though he acknowledged the test was limited to a single repository. "This is just one source code repository and maybe it is much better on other things. I can only tell and comment on what it found here," he cautioned.

AI Code Analysis: A Sobering Reality Check

Despite his criticism of Mythos's marketing, Stenberg reiterated his broader belief that AI-powered code analyzers are fundamentally superior to traditional static analysis tools. "AI powered code analyzers are significantly better at finding security flaws and mistakes in source code than any traditional code analyzers did in the past," he emphasized. He noted that all modern AI models are now effective at this task, democratizing vulnerability discovery. "Anyone with time and some experimental spirits can find security problems now. The high quality chaos is real."

Stenberg's critique underscores a growing tension in the security community: the gap between vendor claims and real-world utility. While AI tools have improved, the notion that any single model represents a quantum leap may be overblown. His words offer a needed counterpoint to the hype cycle surrounding large language models in security contexts.

Background: The Mythos Controversy

Anthropic, an AI safety startup, announced Mythos earlier this year, describing it as a cutting-edge code analysis system capable of autonomously finding critical vulnerabilities. The company withheld the model from public release, citing fears that it could be weaponized. This decision sparked intense debate in cybersecurity circles about responsible disclosure and the potential risks of advanced AI.

Stenberg, who has maintained the curl project for over two decades, decided to test Mythos personally by applying it to curl's codebase. His findings directly challenge the premise that Mythos is uniquely powerful. The cloud of secrecy around the model, he implies, may have been a strategic marketing move rather than a genuine safety concern.

What This Means

Stenberg's analysis suggests that enterprises relying on Mythos for security audits should not expect dramatically better results than what they could achieve with other AI code analyzers. The real breakthrough, according to him, is the broader maturation of AI-powered security tools—not any single vendor's offering. For security teams, this means they can choose from a competitive landscape of effective tools rather than pinning hopes on one "dangerous" model.

The broader implication for the AI industry is a call for transparency and rigorous benchmarking. Stenberg's willingness to publicly test a model that was deemed too risky for release sets a precedent for independent evaluation. "The high quality chaos is real" suggests that the power of AI in security is now widely accessible, but it is also democratized—anyone can find flaws. The danger, if any, lies not in a single tool but in the collective capability of many.

As the dust settles, Stenberg's critique may temper expectations around Anthropic's future releases and encourage other developers to conduct their own assessments. For now, the curl creator's message is clear: don't believe the hype. Test for yourself.

curl Creator Dismisses Anthropic's Mythos AI as Overhyped: 'Primarily Marketing'