The U.S. Department of Commerce announced a deal on Tuesday that creates a new class of digital privilege in the heart of Washington. Under the agreement, elite tech firms including Google DeepMind, Microsoft, and xAI will allow the government to review early versions of their powerful AI models before they reach the public. The Center for AI Standards and Innovation (CAISI) stated that this review process is necessary to protect national security from risks involving biosecurity and cybersecurity. Yet, the enclosure of these safety tools for the benefit of enterprise partners leaves the average citizen in the dark.

Inside the conference rooms of CAISI, officials are particularly concerned about models like Anthropic’s Mythos, which experts fear could help hackers exploit vulnerabilities at an unprecedented scale. While the giants of Silicon Valley trade safety protocols with federal bureaucrats, the public software repositories used by small businesses remain unshielded. This is the birth of a cognitive enclosure, where the elite hoard the defenses while the commons are left exposed to the algorithms.

"The review process would be key to understanding the capabilities of new and powerful AI models," CAISI announced. We believe this is a classic example of the hollow state prioritizing the protection of corporate assets over the security of the family home. If these tools are dangerous, they should not be the private property of a few gated hubs. The cold hum of the server farm should not be the only voice heard in the halls of power. A just society requires that the tools of safety belong to the many, not just the few who can afford the subscription.