Jamie Dimon is known for many things: running the biggest bank in the U.S., writing annual letters that double as economic treatises, and occasionally saying something that makes regulators wince. But lately, he's been doing something a bit more discreet: picking up the phone and calling fellow CEOs to talk about artificial intelligence. Not the fun, productivity-boosting kind of AI talk, but the scary, keep-you-up-at-night kind.

According to a report from Reuters on Wednesday, Dimon has been personally contacting the heads of major banks, regional lenders, and IT companies to expand the Alliance for Critical Infrastructure (ACI), a JPMorgan-backed initiative aimed at shoring up the security of critical infrastructure. The group's pitch? Let's get ahead of AI's risks before they get ahead of us.

The ACI and Dimon have already reached out to more than 40 companies across critical infrastructure sectors to discuss cybersecurity collaboration, with calls planned for August. The outreach started in July, but so far, the group hasn't announced any concrete outcomes. Still, the timing makes sense. Recent cyberattacks on U.S. water systems have highlighted just how vulnerable our essential services can be, and the need for better information sharing across industries has never been more obvious.

JPMorgan is a founding member of the ACI, alongside Mastercard Inc. (MA), Berkshire Hathaway Energy of Berkshire Hathaway Inc. (BRK), and others. The group is now working to expand its AI efforts to strengthen coordination and resilience against cyber, physical, and geopolitical threats. In a statement to Reuters, Dimon said the ACI recognized the importance of prioritizing AI years ago and brought critical infrastructure companies together to address the challenge.

This isn't just corporate hand-wringing. AI adoption is accelerating, and with it, concerns that the technology could supercharge cyberattacks and expose financial systems and critical infrastructure to new vulnerabilities. Regulators and industry leaders alike have been sounding the alarm, and Dimon has been one of the more vocal figures.

Back in July, Dimon warned that Anthropic's Mythos AI model poses serious security risks, saying U.S. officials are actively working on the issue. He went so far as to compare broadly distributing such powerful AI to "giving ballistic missiles to individuals." That's a vivid image, and it's clear he's not just being dramatic for effect.

Washington Ramps Up AI Oversight

The U.S. government is also stepping up its game. In July, it launched the Gold Eagle initiative, which aims to help AI developers, critical infrastructure operators, and federal agencies share information on vulnerabilities and coordinate responses to advanced AI risks. It's a recognition that AI threats don't respect traditional boundaries, and neither should the defenses against them.

Last week, President Donald Trump said his administration is considering new safeguards for artificial intelligence after OpenAI disclosed its Hugging Face AI breach incident. While acknowledging the need for stronger oversight, Trump emphasized that any regulations must not slow U.S. innovation or allow China to gain an edge in the global AI race. He said the administration is focused on both safety and maintaining American leadership in AI.

Speaking of China, concerns over Chinese open-weight AI models have been growing. Beijing-based Moonshot AI recently launched its Kimi K3 model, which reportedly matched leading OpenAI and Anthropic systems on some benchmarks. That prompted the Trump administration to renew efforts to restrict foreign open-source AI models, amid allegations that Moonshot AI violated U.S. intellectual property rights.

So, what's the takeaway here? Dimon's quiet campaign is a sign that the financial industry is taking AI risks seriously, not just as a regulatory compliance issue, but as a fundamental threat to the systems we all rely on. The ACI's expansion could be a meaningful step toward building a more coordinated defense. But as with any initiative, the real test will be in the execution. For now, the calls are being made, the meetings are being scheduled, and the message is clear: AI's benefits come with serious risks, and it's better to prepare now than to react later.