Anthropic shelves 'mythos' ai – raising cybersecurity concerns and sparking pr war
San Francisco – ai firm Anthropic has pulled the plug on releasing its groundbreaking new model, ‘Mythos,’ citing ‘overwhelming responsibility,’ a move that’s ignited a flurry of debate, raised cybersecurity alarms, and fueled a high-stakes public relations battle.
Strategic delay sparks industry scrutiny
The decision, announced this week, comes after Treasury Secretary Scott Miller convened a meeting with the heads of major banks to discuss ‘Mythos,’ a model reportedly possessing unprecedented capabilities. Conservative MP Danny Kruger has already demanded government intervention, warning of ‘catastrophic cybersecurity risks’ for the UK, while the tech sector grapples with the implications of Anthropic’s cautious approach.
Openai under scrutiny
The move has also intensified scrutiny of Anthropic’s rival, OpenAI, which recently offered the US military assistance in deploying similar ai tools. Interestingly, Anthropic emerged more favorably in a recent spat with the Department of Defense, despite creating the ai underpinning the Pentagon’s recent strike on Iran – a contrast that’s raised eyebrows and fueled speculation about strategic messaging.
A calculated pr blitz
Anthropic’s leadership, spearheaded by Dario Amodei, has been the subject of a relentless PR campaign in recent months, including a New Yorker profile, articles in the Wall Street Journal, and a prominent feature on the cover of Time magazine. Amodei’s image even graced the cover alongside Defense Secretary Pete Hegseth, a deliberate move to cultivate an image of responsible innovation.
Technical expertise vs. hype
Despite Amodei’s claims of technical superiority over OpenAI’s Sam Altman – who, according to one AI critic, “seems to have graduated from the same school of hype and exaggeration” – questions linger. The accidental release of a portion of ‘Mythos’’s source code earlier this month, followed by claims of stewardship over emerging cyber threats, has prompted skepticism.
Limited capacity and strategic signaling
Analysts suggest Anthropic’s decision may be rooted in resource constraints, with the company reportedly struggling to meet demand for its Claude model, implementing usage caps and requiring users to purchase additional capacity. This raises the possibility that ‘Mythos’’s release was strategically timed to showcase Anthropic’s capabilities while simultaneously limiting independent assessment. Danielle Ghiglieri, Anthropic’s media lead, defended the company’s approach, highlighting the “mad dash” to secure media coverage and emphasizing the importance of controlled rollout.
A measured response
Jameison O’Reilly, a cybersecurity expert, acknowledged the significance of ‘Mythos’ but cautioned against overblown claims, stating that ‘zero-day vulnerabilities’ found by the company were relatively insignificant in real-world scenarios. He noted that Anthropic’s caution aligns with OpenAI’s own strategy of prioritizing safety through PR, suggesting a calculated attempt to build public trust before pursuing profit-driven expansion. The battle for dominance in the AI landscape is now not just about technological advancement, but about perception and control.
The race for investment continues
Ultimately, Anthropic’s decision to withhold ‘Mythos’ underscores the complexities and risks associated with rapidly advancing AI. Like OpenAI, the company is engaged in a fierce competition for investment and market share, vying to capture the attention and loyalty of users – from those seeking personalized AI companions to businesses looking to automate operations. But despite the hype, the fundamental differences between these models remain marginal, resting largely on intangible qualities like ‘sense of self’ – a distinction that may prove increasingly difficult to quantify.