UK regulators are assessing risks from Anthropic's AI model, Claude Mythos, amid cybersecurity concerns and significant discoveries of vulnerabilities.
UK financial regulators are actively engaging with the government’s cybersecurity agency and major banks to assess the risks posed by Anthropic's new AI model, Claude Mythos, according to reports. The Bank of England, the Financial Conduct Authority, and the Treasury are collaborating with the National Cyber Security Centre to investigate vulnerabilities in critical IT systems identified by the AI.
In the coming weeks, significant financial institutions, insurance companies, and exchanges will be briefed on cybersecurity risks associated with Mythos. This follows the alarming discovery of thousands of zero-day vulnerabilities in major operating systems and web browsers by the AI within a short timeframe.
AI Model's Limited Release and Testing
Due to its powerful capabilities, Anthropic opted not to release Mythos publicly. Instead, the company has initiated Project Glasswing, a controlled testing environment involving industry giants such as Google, Microsoft, and JPMorgan Chase.
U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an emergency meeting with Wall Street executives to discuss the implications of Mythos, emphasizing that its launch signifies a new chapter in cybersecurity. Canadian financial authorities are also assessing systemic risks linked to this advanced AI model.
Marketing or Genuine Caution?
Critics suggest that restricting access to Mythos could be a strategic marketing move. Dan Lahav, head of AI startup Irregular, pointed out that while automated vulnerability scanning is essential, the actual risk posed by identified flaws can vary significantly.
Despite Anthropic’s claims that Mythos can exploit vulnerabilities more effectively than its predecessor Opus, the true efficacy of the model in cybersecurity remains uncertain. The startup Aisle has reported replicating many of Mythos's capabilities using different neural networks, indicating that no single model is universally applicable to cybersecurity needs.
David Crawshaw, CEO of exe.dev, remarked that the limited availability of top models is a strategy to secure exclusive contracts with large enterprises. He noted that when Mythos becomes publicly available, it may lead to the emergence of a new advanced model tailored for corporate clients.
Anthropic's Rapid Growth
Anthropic has seen substantial growth in recent months, with increasing demand for its solutions. At the HumanX AI conference in San Francisco, Claude was highlighted as one of the leading products in the market, overshadowing competitors like ChatGPT.
Despite some analysts claiming that OpenAI has lost its competitive edge, a recent financial analysis indicates that both OpenAI and Anthropic are among the fastest-growing companies in history. OpenAI is responding to the competition by introducing a new $100 subscription plan for broader access to its programming tool, Codex, while Anthropic continues to innovate with offerings like Claude Managed Agents.
