The White House has formally launched a cybersecurity coordination group that pulls together frontier AI developers and operators of U.S. critical services to trade information on software vulnerabilities that advanced models are turning up, according to a statement released this week. It fulfills the executive order President Donald Trump signed in June and slots into a growing federal role in monitoring how AI models are used and abused.
Anthropic and OpenAI at the table
Systems from labs including Anthropic and OpenAI have gotten sharp enough to identify software and infrastructure flaws at scale, and U.S. officials worry that adversaries can use them to hunt weaknesses in the systems underpinning banks, hospitals and power grids. The coordination group is meant to short-circuit that: developers and essential-services operators share what they find, avoid duplicating work and coordinate remediation before flaws get burned.
Open-source models are in the tent too
Cyber director Sean Cairncross said the arrangement includes developers of open-source AI models, though he did not identify which. In the U.S. that pool includes Nvidia, Meta Platforms and Reflection AI. The inclusion of open-weight developers matters because their models can be repurposed offline and can't be centrally patched — once a capability is out, it is out — which puts a premium on getting critical-infrastructure operators aligned on where the weaknesses actually are.
Trump's shifting AI posture
The move continues a pivot inside the Trump administration. At the start of the second term the White House signaled a light-touch stance on AI. Since then it has taken a more active role — establishing the early-access review process for frontier models, driving reciprocal moves with South Korea's $880B AI push, and now standing up direct coordination between labs and grid, hospital and financial-infrastructure operators.
Where the risk actually lives
The setup lands in the middle of a busy summer for AI-security disclosures. In the past week, WordPress force-patched a pre-auth RCE affecting every 6.9 and 7.0 site, and Microsoft floated its own multi-model cybersecurity product, Project Perception, that will route work between Anthropic, OpenAI and its own models. Those signals — and the growing frequency of ransomware hits like the Coca-Cola Fairlife dairy shutdown — are why the White House wants a formal channel between the model builders and the operators actually running exposed infrastructure. The order directed the Treasury Department, National Cyber Director's Office, Department of Defense and NSA to build the collaboration; the operational plumbing is still being wired in.
Reporting based on coverage from the White House, Reuters, and Claims Journal.