US tech firm’s new AI model sparks global concern but may be overhyped, says cybersecurity professor

The US-based Anthropic company has warned that its new Mythos AI model poses unprecedented cybersecurity risks, though experts suggest the claims may be exaggerated, reports Finnish broadcaster Yle.

Anthropic has delayed the public release of Mythos, citing its ability to identify and exploit software vulnerabilities at an alarming level. The company claims the model could threaten critical infrastructure if misused, prompting high-level discussions with US authorities, including Treasury officials and financial regulators.

However, Jyväskylä University cybersecurity professor Markus Miettinen downplays the revolutionary nature of the technology. “Based on the limited information released so far, this appears to be another step in an ongoing evolution rather than a breakthrough,” he told Yle. The company has promised a detailed technical report in July, which may clarify its capabilities.

Miettinen speculates that Anthropic’s dramatic warnings could be a calculated marketing strategy. “Major tech firms naturally want early, exclusive access—but I doubt Mythos is so dangerous it can’t be shared. This is likely more about hype than a genuine revolution.”

Global interest and cautious optimism
The model has attracted unusual attention worldwide, with only select partners like Amazon, Apple, and Microsoft granted initial access. Matias Mesiä, operations chief at Finland’s National Cyber Security Centre, acknowledges Mythos may outperform existing tools in vulnerability detection. “As AI advances, both attackers and defenders gain new capabilities,” he noted.

For consumers, Mesiä advises prioritising devices with long-term security updates. “Avoid outdated phone models—ensure at least three years of remaining support to mitigate risks from evolving AI-driven threats.”