CL-SN55 941 —
GPT-61S 885 —
GPT-56T 861 —
GPT-6S 825 —
CL-OP55X 818 -0.1%
MUSE-SPK 793 —
QWEN-38X 779 —
GPT-6A 778 —
CL-OP5H 711 -0.1%
CL-OP5X 709 -0.1%
KIMI-K3X 699 -0.1%
CL-OP46H 694 -0.1%
CL-FAB5H 692 -0.1%
CL-OP55H 692 -0.1%
CL-OP47H 683 -0.1%
GEM-37FH 648 -0.2%
GEM-38FH 633 -0.2%
INKL 531 —
CL-OP47 529 -0.2%
GPT-56S 529 -5.9%
GEM-3P 493 —
CL-OP46 488 —
CL-OP48 481 —
GEM-31P 453 -0.2%
GPT-6L 436 —
CL-SN55 941 —
GPT-61S 885 —
GPT-56T 861 —
GPT-6S 825 —
CL-OP55X 818 -0.1%
MUSE-SPK 793 —
QWEN-38X 779 —
GPT-6A 778 —
CL-OP5H 711 -0.1%
CL-OP5X 709 -0.1%
KIMI-K3X 699 -0.1%
CL-OP46H 694 -0.1%
CL-FAB5H 692 -0.1%
CL-OP55H 692 -0.1%
CL-OP47H 683 -0.1%
GEM-37FH 648 -0.2%
GEM-38FH 633 -0.2%
INKL 531 —
CL-OP47 529 -0.2%
GPT-56S 529 -5.9%
GEM-3P 493 —
CL-OP46 488 —
CL-OP48 481 —
GEM-31P 453 -0.2%
GPT-6L 436 —
← Back to feed

Bank of England, Fed, and Treasury Convene Emergency Sessions Over Claude Mythos — White House Extends Access to Federal Agencies

The release of Claude Mythos Preview has triggered the most coordinated multilateral regulatory response to an AI model in history. Finance ministers, central bankers, and financial supervisors across the US, UK, Europe, and Japan convened emergency sessions this week after evaluations confirmed the model can autonomously identify and exploit software vulnerabilities at a speed and scale that exceeds what human red teams can defend against in real time.

What Regulators Are Saying

Bank of England Governor Andrew Bailey told the BBC directly: “We are having to take this very seriously.” The statement is notable — central bank governors do not typically comment publicly on AI model releases. The UK’s Cross Market Operational Resilience Group (CMorg), chaired by the Bank of England, has been coordinating a briefing to every major British financial institution.

In the US, Treasury Secretary Scott Bessent and Fed Chair Jerome Powell convened a closed-door session with chief executives of the largest American banks. The explicit subject: the systemic risk posed by Mythos-class AI to financial infrastructure. JPMorgan Chase, Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley are among the banks that have begun in-house testing of the model. The White House is encouraging this — and, as of Thursday, Bloomberg reports the US government plans to extend access to a version of Mythos to major federal agencies.

The European Central Bank has also entered the picture, asking banks to report their risk exposure to Anthropic’s model. German banking supervisors are running joint assessments with cybersecurity agencies. Japan’s regulators are separately evaluating implications for Japanese banking and critical infrastructure.

What the UK AI Security Institute Found

The only independent evaluation of Mythos comes from the UK AI Security Institute, published April 14. Key findings:

  • 73% success rate on expert-level capture-the-flag (CTF) cybersecurity tasks — tasks that typically require years of specialist experience
  • First AI model to complete a full 32-step simulated enterprise network attack autonomously
  • Confirmed ability to chain previously unknown vulnerabilities across large codebases

The AISI report was careful: “Our testing shows that Mythos Preview can exploit systems with weak security posture, and it is likely that more models with these capabilities will be developed.” The report also noted Mythos is not dramatically better than Opus 4.6 — but the capability threshold it crosses is qualitatively different.

Anthropic’s Response: Controlled Access and a Safer Surrogate

Anthropic has kept Mythos restricted to 40 organisations through Project Glasswing, all of them defensive partners — AWS, Apple, Cisco, CrowdStrike, Google, JPMorgan Chase, Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks among them. No public access. No API tier.

On April 16, Anthropic released Claude Opus 4.7 as a deliberate intermediate step: a model with reduced cyber capabilities relative to Mythos, but shipping with automated safeguards that detect and block prohibited cybersecurity requests. The stated goal is to test the safeguard infrastructure at scale before any broader Mythos release.

Anthropic CEO Dario Amodei is reported to be meeting White House Chief of Staff Susie Wiles on Friday, indicating a potential breakthrough in an ongoing dispute with the Pentagon over AI policy.

The Deeper Issue

The regulatory response reflects a structural problem that Mythos has made concrete. Financial institutions — and critical infrastructure more broadly — run software stacks that include decades of legacy code, undiscovered vulnerabilities, and tight interdependencies. A model that can systematically scan large codebases and chain previously unknown flaws changes the attack surface calculation for every sector that has not kept pace with patching.

The same training dynamics that produced Mythos’s emergent cybersecurity capabilities will produce equivalent capabilities in models from other labs. The AISI report explicitly flagged this. The question regulators are now racing to answer is not specific to Anthropic — it is whether defensive deployment can outpace the proliferation of models with equivalent capability.

Anthropic has not announced a timeline for broader Mythos access.