Anthropic To Brief Global Financial Watchdog On Cyber Flaws Exposed By Mythos

Anthropic is reportedly preparing to brief the Financial Stability Board (FSB) on cyber vulnerabilities in the global financial system identified by its newest AI model, Claude Mythos Preview. The move signals rising regulatory concern over how frontier AI could reshape both cyber defense and cyberattack risks.

 

TL;DR

• Anthropic is expected to brief the FSB on cyber risks linked to Claude Mythos Preview.
• The briefing was reportedly requested by Bank of England Governor Andrew Bailey, who also chairs the FSB.
• Mythos can identify and help fix serious software vulnerabilities, but the same capability could be misused by attackers.
• The model is currently limited to selected partners through Anthropic’s Project Glasswing.

 

What Happened?

Artificial intelligence startup Anthropic is set to discuss cybersecurity vulnerabilities in the global financial system with members of the Financial Stability Board, according to a Financial Times report cited by Reuters. The briefing will focus on flaws identified by Anthropic’s latest AI model, Claude Mythos Preview.

Source

The FSB, which coordinates financial regulation across G20 economies, is reportedly seeking a closer look at the potential risks after Mythos exposed weaknesses across critical software, infrastructure, and web systems. The meeting was requested by Bank of England Governor Andrew Bailey, who also chairs the FSB.

Reuters reported the development on May 18, 2026, while noting that it could not immediately verify the Financial Times report independently. Neither Anthropic nor the FSB had publicly commented on the report at the time of publication.

 

Why Mythos Has Caught Regulators’ Attention?

Claude Mythos Preview is not a regular AI chatbot rollout. Anthropic says the model is being made available in a gated research preview through Project Glasswing, an initiative designed to help major technology firms, banks, open-source maintainers, and infrastructure providers find and fix vulnerabilities before attackers exploit them.

Anthropic said Project Glasswing launch partners include Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. The company has also extended access to more than 40 additional organizations that build or maintain critical software infrastructure.

The concern is that Mythos-class systems could speed up both sides of the cybersecurity race. For defenders, such models can uncover hidden vulnerabilities, support patching, and test systems at a scale that human teams cannot easily match. For attackers, similar capabilities could help discover zero-day flaws faster, chain weaknesses together, and automate more sophisticated intrusions.

Financial Sector Risk Comes Into Focus

 

The financial industry is especially exposed because banks, payment providers, trading systems, and market infrastructure often depend on shared software vendors, common cloud providers, and legacy systems. A vulnerability found in widely used infrastructure could therefore create simultaneous risks across multiple institutions.

The IMF has also warned that AI-powered cyber tools could become a financial stability concern, especially when attackers can operate faster than institutions can patch and remediate vulnerabilities. It noted that shared service providers and common software can create correlated risks across the financial system.

Reuters separately reported that the Bank of Spain recently called for broader access to protective AI tools and stronger international cooperation, citing cyber risks from models such as Mythos. The central bank said advanced AI could accelerate the exploitation of software vulnerabilities and potentially threaten global financial stability.

 

Anthropic’s Defensive Push

 

Anthropic’s public position is that Mythos is being kept in limited circulation to prioritize defensive use. The company says it does not plan to make Claude Mythos Preview generally available and wants to improve safeguards before broader deployment of similar capabilities.

Through Project Glasswing, Anthropic has committed up to $100 million in model usage credits and $4 million in donations to open-source security organizations. The company says partners will use Mythos Preview for tasks such as vulnerability detection, black-box testing, endpoint security, and penetration testing.

What’s Next?

 

The reported FSB briefing could become an important test case for how global financial regulators respond to frontier AI systems with dual-use cyber capabilities. The question is no longer whether AI can help find vulnerabilities, but how quickly regulators, banks, and software providers can turn those findings into safer systems.

For Anthropic, the challenge will be balancing responsible access with the need to give defenders the same level of AI capability that attackers may eventually obtain. For financial institutions, the message is clearer: patching, resilience testing, vendor risk management, and AI-assisted defense are becoming board-level priorities.