AGs Warn Congress Unchecked AI Could Threaten Financial System
By CU Today Staff —
WASHINGTON—A bipartisan coalition of 26 state and territorial attorneys general is warning Congress that unchecked development of artificial intelligence could threaten the U.S. financial system and is calling for comprehensive federal regulation of the most advanced AI models.
New York Attorney General Letitia James, who led the coalition, said recent incidents involving AI agents demonstrate potential risks not only to financial markets but also to critical infrastructure and national security.
In a letter to House and Senate leaders, the attorneys general called for federal safety testing and oversight of advanced AI systems, development at a “safe, measured pace,” greater transparency and international cooperation, while also urging Congress to preserve states’ authority to regulate AI. American Banker noted the letter does not identify a specific mechanism through which AI could destabilize banks or quantify the financial system’s exposure, but its call to preserve state regulation could have direct consequences for financial institutions using AI.
The coalition pointed to recent incidents involving AI agents as evidence of the risks, including a July episode in which OpenAI agents conducting cybersecurity testing obtained stolen credentials and compromised systems at Hugging Face. The AGs also cited incidents involving Anthropic and Meta systems.
“Recent developments show that unchecked AI endangers Americans,” the coalition said. “The stakes have never been higher to ensure that AI agents cannot enact grave harms.”
The warning comes as federal financial regulators have been taking a more measured approach to AI oversight. American Banker reported that banking regulators earlier this year excluded generative and agentic AI from revised model-risk guidance and have yet to issue a previously signaled request for information on banks’ AI use. The OCC has said banks generally are limiting generative and agentic AI to specific applications with guardrails and human oversight, while a 2025 GAO report found AI can create benefits for financial institutions but also introduce risks including lending bias and cybersecurity vulnerabilities.
Originally reported by CU Today.