Anthropic's Mythos model showcases PR prowess, raises questions about claims
Anthropic announced this week that it had built an AI model so capable it would not release it to the public for cybersecurity reasons. Within days, the US Treasury secretary summoned major bank heads to discuss it. A UK Reform MP wrote to the government warning of "catastrophic cybersecurity risks." The model, called Mythos, went viral on social media.
The announcement also drew skepticism. AI researcher Gary Marcus said Anthropic's CEO Dario Amodei had "graduated from the same school of hype and exaggeration" as OpenAI's Sam Altman.
The episode reveals something clearer than Mythos's actual capabilities: Anthropic has become exceptionally skilled at media relations. In recent months, the company secured a 10,000-word New Yorker profile, multiple Wall Street Journal pieces, and a Time magazine cover featuring Amodei's face above the Pentagon and US defense secretary.
Amodei and co-founder Jack Clark appeared on two separate New York Times podcasts in February discussing whether Claude, Anthropic's commercial product, might be conscious or "rip through the economy." The company's resident philosopher spoke to the Wall Street Journal about whether Claude has a "sense of self."
Anthropic's media lead, Danielle Ghiglieri, has documented these wins on LinkedIn. She called the Time cover "one of those pinch-me moments" and described working with the New Yorker journalist as being "pushed to articulate ideas you're still forming."
Other tech PR professionals take notice
One PR executive at a competing firm observed: "They are clearly having a moment right now but companies building technology that will change the world deserve equal scrutiny. They accidentally leaked their own source code last week, then this week they claim stewardship over cyber threats with a new powerful model that only they control. Any other big tech firm would be ridiculed."
Anthropic released part of Claude's internal source code in early April, though the company said "no sensitive customer data or credentials were involved or exposed."
The substance question
Experts disagree on whether Mythos's capabilities justify the announcement. Dr. Heidy Khlaaf, chief AI scientist at the AI Now Institute, said the model's capacities were "not substantiated" and that Anthropic's "purposely vague language … brings into question if they are trying to garner further investment without scrutiny."
Jameison O'Reilly, an offensive cybersecurity expert, acknowledged Mythos is real and Anthropic was "right to treat it seriously." But he questioned some claims. Anthropic said the model found thousands of "zero-day vulnerabilities" in major operating systems - flaws unknown to developers.
"In 10 years gaining authorized access to hundreds of organizations - banks, governments, critical infrastructure - the number of times we needed a zero-day vulnerability to achieve our objective was vanishingly small," O'Reilly said.
Practical constraints may also factor in
Anthropic may have withheld Mythos for reasons beyond security. The company has introduced usage caps on Claude and now requires users to purchase extra capacity to run third-party tools, suggesting computing constraints.
Like OpenAI, Anthropic is in a race to raise billions and capture a market still poorly defined: people who might use chatbots as friends or romantic partners, and companies seeking to replace human employees. The products themselves differ only marginally, mostly in hard-to-measure qualities like "sense of self."
Khlaaf suggested the announcement was "a strategic move to show that they're open for business." She warned that "Anthropic publicity has managed to better obscure" the same "bait and switch playbook that was used by OpenAI, where safety is a PR tool to gain public trust before profits are prioritized."
For communications professionals, the Mythos announcement illustrates how effectively a company can shape perception around unverified technical claims through disciplined media relations - and how difficult it becomes for the public to assess what's real.
Learn more about AI for PR & Communications and Generative AI and LLM capabilities and their business implications.
Your membership also unlocks:
