Switzerland's financial supervisory authority has signaled serious concern about the implications of widespread access to Anthropic PBC's artificial intelligence model Mythos, saying that unfettered availability could amount to a systemic risk to the national financial system.
"The uncontrolled and immediate availability of AI models such as Mythos would be classified as a systemic risk," a Finma spokesperson said, explaining that a large-scale release could expose virtually all existing software to a wave of previously unknown zero-day vulnerabilities. Finma added that those vulnerabilities could be exploited quickly and with the aid of AI.
Anthropic has publicly described Mythos as a potent tool for uncovering software vulnerabilities and has said it is too powerful to offer to the general public. The company has stated it will grant access only to a limited set of carefully selected parties. Anthropic has also warned that if such tools were to fall into malicious hands, they could become a potent means for attackers to steal data or impair critical infrastructure.
Regulators, central bankers and corporate leaders are actively seeking greater clarity on the technology and how it should be governed. There are explicit concerns noted by officials that financial systems outside the United States, including those in Europe, may be disadvantaged if they lack sufficient access to tools such as Mythos.
Bundesbank President Joachim Nagel captured that tension when he said on Tuesday: "We must prevent the misuse of this technology. At the same time, all relevant institutions should have access to such technology to avoid competitive distortions."
Finma emphasized that it is treating the swift evolution of artificial intelligence seriously and is engaged with national cybersecurity authorities, banks and providers of critical services. The regulator said it is also coordinating with international counterparts.
On the operational front, Finma urged banks to integrate the changing threat environment into their risk-management frameworks. "Cyber attacks are becoming faster, more precise, and easier to carry out with the help of AI," the regulator warned, underlining the need for financial institutions to adapt controls and monitoring accordingly.
The comments from Swiss authorities reflect a broader dilemma: how to restrict harmful uses of advanced AI while ensuring regulators, critical-service operators and financial institutions have sufficient access to defensive capabilities and intelligence to manage evolving threats.
Clear summary: Finma has warned that unfettered access to Anthropic's Mythos could create a systemic risk for Switzerland's financial system by enabling rapid exploitation of unknown software vulnerabilities. Anthropic says Mythos will be tightly restricted because of its potency, while regulators and central bankers debate access versus misuse.