Anthropic to Brief G20 Regulators on AI Cybersecurity Risks
Anthropic will brief the Financial Stability Board and G20 regulators on cybersecurity vulnerabilities identified by its new AI model, Mythos. Prompted by UK and global concerns, the briefing addresses fears that advanced AI could expose critical weaknesses in financial infrastructure and amplify cyber threats. While Mythos has detected thousands of high-severity flaws, access remains restricted to select firms under Project Glasswing. This coordinated international response aims to mitigate systemic risks to global financial stability and establish best practices for AI adoption in the financial sector.
Editorial summary awaiting refresh
Cross-source coverage
Wire timeline
Bank of Italy Engages AI Firms on Security Risks for Banks
The Bank of Italy is in discussions with global artificial intelligence providers, including Anthropic, ahead of releasing new AI models to the financial sector, Governor Fabio Panetta announced in his annual keynote speech. The central bank recently initiated talks with national authorities, financial players, and IT service providers regarding AI-related risks. Supervisory concerns have focused on Anthropic's Mythos model, which is designed to find flaws in computer code to bolster cybersecurity defenses but is seen by experts as capable of turbo-charging attacks on banks' technology systems. Panetta emphasized that banks are responsible for ensuring their systems' protection and continuity, and third-party providers bear the same responsibility. He urged banks to keep IT risks in check, draw up rapid intervention plans, and use part of their record profits to support necessary investments. Panetta stressed that the response cannot be purely technological but requires sound governance, clear responsibilities, and prompt action plans.
Yahoo FinanceECB convening bank meeting to fix flaws exposed by AI models: report
The European Central Bank (ECB) is planning to convene a meeting with lenders to pressure them into accelerating efforts to secure their IT systems, following vulnerabilities exposed by advanced AI models. According to a report, the AI model 'Mythos', developed by Anthropic, has identified unknown flaws in banking IT systems, causing widespread anxiety across Europe. Anthropic is reportedly limiting access to Mythos due to the sensitive nature of the findings. The ECB's initiative aims to address these systemic risks and push banks to strengthen their cybersecurity infrastructure. The meeting underscores growing regulatory concern over the intersection of artificial intelligence and financial sector stability.
The Business TimesECB convening bank meeting to fix flaws exposed by AI models: report
The European Central Bank (ECB) is planning to urge lenders to accelerate efforts to secure their IT systems after organizing a meeting to address vulnerabilities exposed by artificial intelligence models. According to a report, the AI tool named Mythos, which can identify unknown flaws in IT systems, has caused a wave of anxiety across Europe. Anthropic, the developer of Mythos, is limiting access to the tool. The ECB's initiative aims to pressure banks to fix these security gaps more quickly, highlighting growing concerns over AI-discovered vulnerabilities in the financial sector's technological infrastructure.
The Business TimesECB Calls Emergency Meeting with Banks Over AI-Driven Cybersecurity Threats from Anthropic's Mythos
The European Central Bank (ECB) is convening European banks on Tuesday to address escalating cybersecurity risks posed by advanced AI models, particularly Anthropic's Claude Mythos Preview. ECB Executive Board member Frank Elderson warned that banks must accelerate patching because AI can now reverse-engineer software fixes within minutes. Mythos has identified thousands of zero-day vulnerabilities across major operating systems and browsers, producing working exploits on its first attempt over 83% of the time. Only 40-50 organizations, mostly US firms like Amazon and JPMorgan Chase, have access through Anthropic's Project Glasswing; no European bank is included. The access gap has stalled EU negotiations with Anthropic, prompting French AI startup Mistral AI to develop a rival cybersecurity model for European banks. Real-world data shows advanced AI models are discovering vulnerabilities at seven times the usual rate, with industry warning of only three to five months of defensive capability remaining.
The Next WebUK Regulators Urge Firms to Mitigate Risks From Frontier AI Models
British financial authorities, including the finance ministry, the Bank of England (BoE), and the Financial Conduct Authority (FCA), have issued a joint warning urging UK companies to proactively plan for and mitigate risks associated with new frontier artificial intelligence models. In a statement released on Friday, the regulators highlighted that the cyber capabilities of current AI systems already surpass those of skilled human practitioners in terms of speed, scale, and cost-efficiency. They warned that if these capabilities are exploited maliciously, they could significantly amplify cyber threats, endangering firm safety, customer security, market integrity, and overall financial stability. This advisory follows recent comments by BoE Governor Andrew Bailey, who identified major cybersecurity risks linked to Anthropic’s Mythos AI product. Cyber experts have previously cautioned that such advanced AI tools could supercharge complex cyberattacks, posing substantial challenges to the banking industry's existing technological defenses. The guidance underscores the urgent need for British firms to adapt their risk management strategies to address the evolving threat landscape posed by rapid advancements in generative AI technology.
Insurance JournalAnthropic to Brief Global Regulators on Mythos AI Cybersecurity Risks
Anthropic is preparing to brief the Financial Stability Board (FSB) on significant cybersecurity vulnerabilities identified by its unreleased AI model, Mythos. The briefing, requested by FSB Chair and Bank of England Governor Andrew Bailey, will be presented to G20 finance ministries and central banks. Bailey previously highlighted Mythos as a critical risk factor, noting its ability to uncover exploitable flaws in major operating systems and browsers with high success rates. This event marks a coordinated global regulatory response, following separate briefings for UK banks, US financial institutions, and regulators in Australia, Europe, and Japan. While approximately 40 to 50 organizations, including major tech firms and JPMorgan, have early access via 'Project Glasswing,' broader regulatory access remains a contentious issue. The briefing aims to address the practical implications of these findings for global financial stability, particularly concerning the balance between offensive cyber capabilities and defensive measures. The situation is further complicated by ongoing political negotiations between Anthropic and the US government regarding export controls and military access to the technology.
The Next WebG20 Seeks Answers on Claude Mythos Risks as Anthropic Prepares Briefing
Anthropic has agreed to brief G20 finance ministries and central banks on cybersecurity vulnerabilities identified by its new AI model, Claude Mythos. This development follows concerns from global regulators, initiated by Financial Stability Board (FSB) chair Andrew Bailey, that advanced AI systems could expose critical weaknesses in the global financial infrastructure. Regulators fear malicious actors might exploit these flaws before they are patched, prompting urges for strengthened cybersecurity and faster software updates. The FSB is preparing a report on sound AI adoption practices for the financial sector. India has also responded, with Finance Minister Nirmala Sitharaman holding high-level meetings to establish real-time threat intelligence-sharing mechanisms. Reports indicate Mythos has already uncovered thousands of high-severity vulnerabilities, including a recent exploit against Apple’s M5 chip protections. Due to these risks, Anthropic has restricted access to the model through Project Glasswing, limiting it to approximately 40 organizations, primarily in the US, following a White House request to prevent broader distribution until safety measures are ensured.
India Today | Latest StoriesBank of England Governor Pushes for Anthropic Briefing on AI Tool Mythos
Bank of England Governor Andrew Bailey has initiated a push for AI developer Anthropic to brief the Financial Stability Board (FSB) on its cybersecurity tool, Mythos. The FSB, which coordinates financial regulation among G20 nations, will review vulnerabilities in the global financial system identified by the tool. This move addresses growing concerns among regulators about the risks AI poses to lenders' cyber infrastructure. Anthropic previously reported that Mythos detected thousands of high-severity vulnerabilities across major operating systems and browsers, warning of severe potential fallout for economies and national security. Access to Mythos remains restricted due to misuse risks, with only select entities like Amazon, Microsoft, and JPMorgan granted access. The briefing coincides with the FSB's preparation of a report on best practices for AI adoption in finance. Meanwhile, UK lawmakers have criticized financial institutions for insufficient AI risk management, with Treasury Select Committee Chair Dame Meg Hillier expressing doubt about the system's readiness for a major AI-related incident.
City AMAnthropic to Brief Financial Stability Board on AI-Identified Systemic Risks
Anthropic, a leading US artificial intelligence company, has agreed to provide a briefing to the Financial Stability Board (FSB) regarding vulnerabilities within the global financial system. This development follows a direct request from the Chair of the FSB. The briefing will focus on specific risks and capabilities identified by Mythos, Anthropic's new AI model. According to sources cited by the Financial Times, the tech group will discuss how these AI-driven insights reveal potential weaknesses in financial infrastructure. This engagement highlights the growing intersection between advanced artificial intelligence technologies and global financial regulation. Regulators are increasingly interested in understanding how large language models and other AI systems might impact market stability, either through operational risks or by uncovering hidden systemic fragilities. The move signifies a proactive step by both the private sector and international regulatory bodies to address emerging technological challenges in finance. By sharing findings from Mythos, Anthropic aims to assist the FSB in its mandate to monitor and mitigate risks to global financial stability, marking a significant moment in the dialogue between AI developers and financial overseers.
TechmemeUK Authorities Urge Firms to Mitigate Risks from Frontier AI Models
The UK finance ministry, the Bank of England, and the Financial Conduct Authority have jointly advised British companies to implement measures to mitigate risks associated with new frontier artificial intelligence models. In a statement released on May 15, 2026, the regulators warned that the cyber capabilities of current AI systems now surpass those of skilled human practitioners in terms of speed, scale, and cost efficiency. They highlighted that malicious use of these technologies could significantly amplify cyber threats, endangering firm safety, customer security, market integrity, and overall financial stability. This advisory follows recent concerns raised by Bank of England Governor Andrew Bailey regarding Anthropic's 'Mythos' product. Cyber experts have previously warned that Mythos has the potential to supercharge complex cyberattacks, posing substantial challenges to the banking industry's existing technological defenses. The joint statement underscores the urgent need for proactive planning and risk management strategies within the financial sector to address the evolving landscape of AI-driven cybersecurity threats.
Latest News