US Treasury and the Fed warn Wall Street: Anthropic’s new AI model could unleash a cyber wave—how big is the risk?

On April 10, 2026, U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened senior Wall Street executives to warn banks about cyber risks tied to Anthropic’s newly released AI model. According to Bloomberg reporting cited by multiple outlets, the meeting focused on how advanced AI tools could accelerate a new wave of cyber threats against financial institutions. The Financial Times adds that Bessent directly called in U.S. bank CEOs, framing the discussion around the model’s ability to detect decades-old vulnerabilities. The reporting also notes that the U.S. government is treating the issue as a systemic financial risk, not a narrow IT problem. Strategically, the episode highlights the growing intersection of AI deployment, cyber offense/defense, and financial stability. The U.S. is effectively signaling to the banking sector that AI-enabled automation can lower the cost and speed of exploitation, raising tail risks for cyber incidents, fraud, and operational outages. This matters geopolitically because cyber resilience and trust in financial infrastructure are core elements of national power, and because AI supply chains are increasingly transnational. While the immediate “who benefits” is ambiguous, the likely winners are cybersecurity vendors, incident-response providers, and cloud/compute suppliers that can demonstrate hardened controls; the likely losers are banks that delay upgrades or underestimate AI-driven threat scaling. Market implications are likely to concentrate in AI compute and cyber-risk pricing. CoreWeave’s shares reportedly jumped as a new Anthropic deal underscored the scramble for AI compute capacity, following an expanded arrangement with Meta announced the previous Thursday, implying continued demand pressure across data centers and GPU supply. On the risk side, the Treasury/Fed warning can raise compliance and security capex expectations for banks, potentially affecting credit spreads for cyber-exposed lenders and increasing demand for cyber insurance. Instruments that may react include bank equities and cyber-defense-related equities, alongside AI infrastructure names; however, the articles do not provide quantified price moves beyond CoreWeave’s stock pop. What to watch next is whether regulators translate the warning into enforceable guidance, supervisory exams, or stress-test scenarios that explicitly model AI-assisted cyber escalation. Key indicators include bank disclosures on security controls, changes in incident-response staffing, and any tightening of vendor risk management for AI tooling. Another near-term signal will be compute contracting behavior—if Anthropic-related demand accelerates, it could further intensify the AI infrastructure cycle that CoreWeave is riding. Escalation triggers would be a major bank cyber incident, evidence of AI-assisted exploitation in the wild, or new guidance from the Treasury/Fed; de-escalation would come from credible mitigation benchmarks and fewer reported breaches tied to AI-enabled attack paths.

Geopolitical Implications

• 01

  AI-driven cyber capability is becoming a cross-border strategic concern, affecting trust in financial infrastructure and national resilience.

• 02

  U.S. regulators are effectively setting a compliance and risk-management benchmark that could influence global banking standards for AI tooling.

• 03

  Compute competition for frontier AI models reinforces the strategic leverage of data-center capacity and supply-chain control in the AI era.

Key Signals

• —Any Treasury/Fed follow-on guidance, supervisory exam themes, or stress-test language referencing AI-assisted cyber escalation.
• —Bank disclosures and internal control updates: model governance, red-teaming, and vendor risk management for AI-related tooling.
• —Cyber incident reporting patterns—especially evidence of AI-assisted exploitation of long-known vulnerabilities.
• —AI compute contracting and capacity announcements tied to Anthropic workloads and related partnerships.