AI model security concerns

What happened: In April, U.S. Federal Reserve and Treasury officials warned top bank CEOs about Anthropic's Mythos AI model, which could pose significant cybersecurity threats. Meanwhile, Chinese AI has matched Mythos in cybersecurity tasks using an open-weight model. Anthropic recently shut down its latest general-use models after a 'jailbreak' vulnerability was discovered in its Fable 5 model. The Five Eyes intelligence alliance also issued a joint statement warning of AI models' increasing cyber risks.

Market impact: This narrative is driving concerns in the financial sector, with banks potentially facing heightened cybersecurity risks. Tech companies developing AI models, like Anthropic, may face regulatory scrutiny and reputational damage. The cybersecurity industry could see increased demand for AI-driven defenses.

What to watch next: The Fed's next monetary policy meeting on June 13-14, where officials may provide further guidance on AI model risks. Anthropic's response to the vulnerabilities and any regulatory actions taken against the company. The release of the Five Eyes' full report on AI cybersecurity risks, expected later this year.