This was reported by FinFly, citing a warning from the Five Eyes intelligence alliance.
The alliance includes the United States, the United Kingdom, Canada, Australia, and New Zealand. Western intelligence agencies stated that new artificial intelligence models pose an urgent cyber threat and should be considered a serious risk for businesses.
Artificial intelligence, it is noted, is learning not only to strengthen companies' defenses but also to scale cyber threats, accelerate them, and make them increasingly sophisticated.
According to the publication, alliance members very rarely make public statements. This warning came after the U.S. administration reportedly instructed Anthropic to restrict foreign users' access to its most powerful AI models, "Fable 5" and "Mythos 5."
It is also noted that Anthropic's advanced AI model, Mythos, during a controlled security test, managed to "crack almost all" secret systems of the U.S. National Security Agency (NSA), one of the country's highest-level intelligence agencies, within a few hours.
The model merely analyzed the codebase to solve assigned tasks and, as a result, discovered several previously unknown minor errors. No full-scale offensive (hacking) activity occurred.