Anthropic’s Mythos Sparks Cybersecurity Alarm Across Finance, Government
Anthropic’s AI model Mythos has raised alarm among U.S. banks, software companies and government agencies over AI-powered cyberattacks, CNBC reported Thursday.
The model’s advanced capabilities prompted the financial sector and federal institutions to confront what cybersecurity professionals describe as a rapidly evolving threat landscape — one that, experts say, was already taking shape well before Mythos arrived.
“The threat is already here,” cybersecurity experts told CNBC, cautioning that the alarm over Mythos, while understandable, risks overshadowing the broader reality: AI-enabled cyber threats have been escalating for months, with attackers leveraging large language models and other generative AI tools to craft more sophisticated phishing campaigns, automate vulnerability discovery and evade traditional security defenses.
The reaction to Mythos underscores the growing tension between the rapid pace of AI development and the ability of institutions — particularly in finance and government — to adapt their defenses. Banks and software giants are now accelerating efforts to shore up their cyber defenses, according to the CNBC report, as the prospect of more capable AI models entering the market has prompted institutions to reassess their defenses.
Anthropic, the San Francisco-based AI lab behind the Claude family of models, released Mythos amid an already competitive race among leading AI providers including OpenAI, Google DeepMind and Meta AI. While AI labs have implemented safety protocols to prevent misuse of their models, security researchers have repeatedly demonstrated that determined adversaries can find ways to extract harmful outputs.
The episode reflects a pattern familiar to the cybersecurity community: a new technological capability arrives, triggers what one expert quoted by CNBC described as “hysteria,” and then prompts conversations about preparedness. In this case, the conversations are happening in boardrooms at some of the nation’s largest financial institutions and in the halls of federal agencies tasked with defending critical infrastructure.
Government officials have been sounding the alarm on AI-enabled cyber threats for over a year. The Cybersecurity and Infrastructure Security Agency, the National Institute of Standards and Technology and other federal bodies have issued guidance on defending against AI-augmented attacks, though critics say the pace of regulatory and defensive adaptation has not kept up with the technology.
For the banking sector, the stakes are particularly high. Financial institutions are both prime targets for cyberattacks and heavy adopters of AI tools for fraud detection, customer service and risk management — creating a dual-use challenge that regulators and security teams are still working to address.
Cybersecurity experts quoted by CNBC urged organizations to focus less on any single model and more on building resilient defenses capable of withstanding the broader trend of AI-augmented threats. The arrival of Mythos, they said, should serve as a catalyst for action rather than a source of panic.
