844 megabytes of federal cloud credentials sat on a public GitHub repository for six months before a security contractor exposed the breach. The leak by the firm Nightwing included plaintext passwords and master keys for AWS GovCloud accounts, essentially surrendering the keys to the U.S. government’s digital kingdom to anyone with an internet connection. This catastrophic failure of the privatized security state has forced the Pentagon to cede what remains of its digital sovereignty to Silicon Valley.

President Trump is now set to sign an executive order anchoring federal cybersecurity in proprietary models from OpenAI and Anthropic. According to Chris Lehane, OpenAI’s Chief Global Affairs Officer, the company is partnering with the White House on a deployment strategy for 'GPT-5.5-Cyber.' Simultaneously, Anthropic’s 'Mythos' model has demonstrated the ability to harvest 10,000 zero-day vulnerabilities, but with a catch: its automated patches often break the underlying code, creating a cycle of dependency.

Smaller towns and local utilities, lacking the personnel to manage these complex AI tools, face the hardest hit. Security officials from over a dozen states have urgently petitioned tech giants for help as they struggle to patch vulnerabilities. This is the Cognitive Enclosure in full effect—a world where the state can no longer defend itself without leasing its survival from tech oligopolies that operate entirely outside public oversight.