The Treasury-led program went live on July 14-15 using Anthropic models and a new VINTS platform to speed vulnerability triage and patching across critical infrastructure and supply chains.
The White House has activated Gold Eagle, the AI-powered cybersecurity initiative ordered under Executive Order 14409 and launched on July 14-15 as a Treasury-led effort backed by agencies including CISA, DHS and the Department of Defense. The program uses Anthropic’s Mythos AI models and a new platform called VINTS, developed with Carnegie Mellon’s Software Engineering Institute, to scan for software vulnerabilities, coordinate reports and accelerate patching across critical infrastructure and supply chains. Gold Eagle is designed to address the backlog and duplication that can slow vulnerability handling across agencies and industries. Officials have framed the initiative as a response to lessons from the 2021 Log4j crisis, when a flaw in widely used open-source software triggered a broad scramble to identify and fix exposures. The government has not disclosed which companies are participating or how the AI prioritizes vulnerability severity. Treasury Secretary Scott Bessent is overseeing the program, with DHS Secretary Markwayne Mullin and Defense Secretary Pete Hegseth in supporting roles. The launch also follows the lifting of export restrictions on certain Anthropic models that had been imposed over cybersecurity concerns. The broader executive order remains notable for barring mandatory licensing or pre-clearance requirements for AI model development while creating a federal mechanism to coordinate AI-assisted cyber defense. Although Gold Eagle is aimed at traditional software infrastructure rather than DeFi or smart contracts, its reach could still matter for crypto firms because exchanges, wallets, bridges and node operators rely on the same open-source and networking components the initiative is intended to protect.