The White House is establishing a new AI Cybersecurity Clearinghouse to help coordinate cybersecurity defenses across critical infrastructure.
Industry and critical infrastructure are racing to keep up with new AI models with increasingly sophisticated capabilities that can discover and exploit (and even protect against) cybersecurity vulnerabilities. Some AI companies are delaying widespread releases of cutting-edge models to give key partners time to patch vulnerabilities before the models are widely available.
The White House Clearinghouse, called Gold Eagle, is a joint project between the Treasury Department, the Department of Homeland Security, and the Department of Defense. AI and cybersecurity companies use the platform to communicate and coordinate their efforts, along with critical infrastructure providers such as utilities and banks.
“These new capabilities will uncover vulnerabilities on a scale we’ve never seen before,” a senior White House official told reporters at a briefing Tuesday.
The goal of the new clearinghouse is to “resolve conflicts, ensure resources are not wasted remediating or scanning the same vulnerabilities, and ensure those vulnerabilities are verified,” the official added. “A team of industry and government engineers then work to triage, prioritize, and remediate those vulnerabilities.”
The White House declined to say which companies are participating in the project, describing them as “open source software partners and U.S. critical infrastructure companies.” The source code of open source software is publicly available and freely available to anyone. View, use, modify, and distribute it. The most well-known AI models from companies like OpenAI, Google, and Anthropic are closed source.
Open source software is ubiquitous, but it’s often run by volunteers who don’t have the money or time to protect their code.
In 2021, a critical bug in open source software left hundreds of millions of devices around the world vulnerable to hacking, prompting a frantic response from Biden administration officials.
The clearinghouse announced Tuesday is a requirement established by an executive order signed by President Donald Trump in June.
The executive order also requires systems that allow AI companies to: Advanced models will be submitted to the federal government for review up to 30 days before being released to other “trusted partners.” This framework needs to be developed by early August, but has not yet been officially established.
but, The White House has already restricted the release of new AI models through other means, such as Anthropic’s export ban, which was eventually lifted. The White House also asked OpenAI to limit the release of its latest models.
The seemingly haphazard approach to regulating the release of new models has led to widespread calls from the AI industry for more consistent regulation.
CNN’s Sean Lyngaas contributed to this report.
