Anthropic Mythos to Release Next Week to Banks in UK

Anthropic plans to release its Mythos AI models to UK-based financial institutions next week, according to Bloomberg.

“The engagement that I’ve had with the CEOs in the last week in the UK has been significant. Customers first want to say that they appreciate our very thoughtful approach to the way in which we have released Mythos,” Pip White, Anthropic’s EMEA North Head, said in an interview. 

“But secondly, they want to understand the opportunity to be considered to gain access to the model in a controlled way.” 

This development comes shortly after Reuters reported that German banks and national authorities are examining risks around Mythos, and that supervisors from the European Central Bank (ECB) are set to quiz bankers about the risks of Mythos.

Mythos, formally introduced as Claude Mythos Preview, is the company’s most advanced frontier AI model to date. 

According to Anthropic’s official disclosures, it is a general-purpose language model that shows a significant leap in overall capability, with particularly strong performance in software engineering and cybersecurity tasks. 

In practice, Anthropic says the model can independently find and exploit zero-day vulnerabilities, chain multiple flaws into full attack paths, and operate across entire software stacks—from auditing codebases to writing and validating fixes—while handling long, multi-step engineering workflows with minimal supervision

The company has highlighted that the model’s ability to identify and analyse vulnerabilities in complex systems emerges from broader advances in reasoning and coding, rather than being narrowly trained for security use cases.

Given these capabilities, Anthropic has opted not to release Mythos broadly. 

Instead, it is being deployed in a highly controlled manner through Project Glasswing, an initiative designed to evaluate and apply the model in defensive cybersecurity contexts. It provides selected organisations with early access to Mythos to help secure critical software systems and prepare industry-wide defences against emerging AI-driven cyber risks. Amazon Web Services, Anthropic, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks are a part of this initiative.

Anthropic says the initiative is intended to strengthen the resilience of global digital infrastructure by enabling partners to discover and fix vulnerabilities before such capabilities become widely accessible, effectively positioning Mythos as both a potential risk factor and a tool for preemptive defence.