Anthropic's Mythos Model Triggers Global Security Scramble
Anthropic has restricted access to Mythos, a powerful artificial intelligence model capable of finding critical vulnerabilities in banking systems, power grids, and government infrastructure. The company granted initial access to only 11 U.S. organizations and Britain, sparking urgent responses from central banks and intelligence agencies worldwide.
The model's cyber capabilities have set off what experts describe as an unprecedented geopolitical competition. The Bank of England governor warned that Anthropic may have "cracked the whole cyber-risk world open." The European Central Bank began quietly surveying banks about their defenses. Canada's finance minister compared the threat to the closure of the Strait of Hormuz.
For government officials, Mythos illustrates a fundamental shift in how advanced AI functions on the global stage. Major breakthroughs now operate less like product launches and more like weapons tests, with nations scrambling to understand the technology and assess security risks.
The Access Question
Anthropic said it is keeping Mythos access limited due to safety concerns. The company has shared the model with more than 40 organizations that maintain critical global infrastructure, including Amazon, Apple, and Microsoft, to help develop security fixes.
Britain's AI Security Institute tested Mythos and published an independent evaluation confirming it could execute complex cyberattacks beyond the capabilities of previous AI models. Kanishka Narayan, Britain's AI minister, said the country was taking steps to protect critical national infrastructure.
The European Commission has met with Anthropic at least three times since the release but has not gained access. Germany's cybersecurity agency also lacks access, though its president met with Anthropic employees in San Francisco for briefings on how the model works.
A Widening Global Divide
Neither China nor Russia has made major public statements about Mythos, but analysts say both countries are watching closely. Inside China, researchers worry the country could fall further behind in the AI race, particularly given U.S. policies restricting Chinese access to advanced semiconductors needed for building AI systems.
Nations without powerful computing infrastructure now face a strategic disadvantage. They risk becoming dependent on U.S. companies while having little influence over how these systems are designed and protected.
Eduardo Levy Yeyati, a former chief economist at Argentina's central bank, said the episode should serve as a policy wake-up call. "Governments can no longer ignore the issue," he said.
Broader Governance Gaps
The scramble over Mythos occurs amid minimal international cooperation on AI. There is no equivalent of the Nuclear Nonproliferation Treaty, no shared inspections, and no agreed-upon rules for handling powerful models like this one.
Anthropic said it expects other groups to release AI models with similar cyber capabilities within 18 months, giving organizations limited time to make necessary security fixes. The company also said it was investigating a report that unauthorized users gained access to a version of Mythos.
On Friday, Dario Amodei, Anthropic's chief executive, met with White House officials after some in the Trump administration noted the potential for the model to damage computer systems. The company said it has no immediate timeline for widely expanding access but will work with the U.S. government and industry partners to determine next steps.
For government workers tasked with cybersecurity or infrastructure protection, understanding these models and their capabilities has become essential. Learn more about AI for Cybersecurity Analysts and AI for Government to stay current on these developments.
Your membership also unlocks:
