After Anthropic Mythos ‘scare’, UK financial services industry regulator wants US-like check on AI models

A top British financial regulator is calling for a review of how powerful artificial intelligence (AI) tools like ChatGPT, Claude and Gemini are managed. The move comes after a recent ‘scare’ involving Anthropic’s advanced Mythos model that raised alarms over how much control AI is gaining over people’s wallets.

According to a report by news agency Reuters, Sheldon Mills, an executive director at the Financial Conduct Authority (FCA), warned on Monday (July 6) that existing rules must evolve quickly. He cautioned that relying too heavily on a tiny handful of tech giants could trigger a massive, system-wide financial meltdown even as the financial watchdog’s urgent review revealed that more than 25% of UK consumers now trust AI chatbots for financial advice.The biggest issue here is that most of these everyday users have no idea that the strict legal protections shielding them when they talk to a human financial advisor do not apply to AI software. If a chatbot gives terrible investment advice that ruins someone financially, the consumer is largely left unprotected.

Frontier models under scanner

Reuters report says that the pressure to act has intensified following a cyber and operational scare involving “frontier” systems such as Anthropic's new Mythos model.

Regulators are also worried about the rise of “agentic systems” that can make automated financial decisions with little to no human supervision.The FCA’s report also highlighted a major structural threat called concentration risk. Mills warned that if the entire banking and financial sector relies on the exact same AI models, cloud servers or tech infrastructure, it creates a dangerous single point of failure.

If one major tech provider crashes or gets hacked, it could cause a domino effect that cripples the entire British financial system at once.To prevent a crisis, Mills recommended that the FCA take action within the next three to six months to decide how to legally bring these general-purpose AI tools under official government oversight.This comes weeks after President Donald Trump signed an executive order asking AI companies to provide models to the federal government to assess their capabilities ahead of a full release. The order asks companies, on a voluntary basis, to participate in a benchmarking process to assess a model’s “advanced cyber capabilities” and determine whether it should be considered a “covered frontier model.”