Anthropic got 90 minutes.
Ninety minutes to pull its most advanced models from every customer, every partner, every researcher relying on them. No specific details about the threat. No transparent process for making the determination. No appeal mechanism. No published standard that Anthropic had failed to meet. Just a directive and a clock.
The Trump administration called it a national security risk after being notified of a jailbreak — a way to get around the model’s internal guardrails. Anthropic said the vulnerability was minor and not unique to their system. Independent cybersecurity researchers who reviewed the actual underlying research agreed with Anthropic’s assessment. Former Facebook chief security officer Alex Stamos reviewed the findings and said there were some valid concerns but nothing that justified a response anywhere near this scale.
The government hasn’t shown its work.
That’s not governance. That’s reach. And the difference between those two things is exactly what the people are going to have to sort out over the next several years.
This wasn’t the first move either. Before the model pulldown, the Department of Defense had already blacklisted Anthropic entirely — labeled them a supply chain risk — after a separate dispute in which Anthropic declined to modify its AI guardrails the way the Pentagon wanted. So the pattern here isn’t a single overreaction to a single incident. It’s a government feeling its way through a domain it doesn’t fully understand, using the tools it has — export controls, blacklists, 90-minute deadlines — because it doesn’t yet have better ones.
That matters. Because the tools you use shape the precedent you set.
When governments act without a transparent, consistent framework — when the process is ad hoc, opaque, and applied unevenly — they don’t just affect the company in the crosshairs. They send a signal to every company watching. They tell researchers, developers, and investors what the rules of engagement look like. And right now that signal is: the rules can change without warning, without explanation, and without recourse.
One legal scholar quoted in the CNN reporting put it plainly. An administration that governs this way will not avoid the heavy regulation it fears. It is manufacturing the conditions for catastrophe or abuse that, in every cycle she had documented, triggers exactly the regulatory overcorrection it was trying to prevent.
That is worth sitting with for a moment.
The argument for light-touch AI regulation — the argument this administration has staked its position on — is that heavy-handed government intervention will slow American AI development and hand the advantage to China. That argument has real merit. The concern is legitimate. Nobody serious thinks the answer is to bury the industry in compliance requirements before the technology is even mature.
But light-touch only works when it is applied consistently and transparently. The moment it becomes selective and opaque, it stops being light-touch. It becomes something closer to arbitrary power. And arbitrary power doesn’t produce innovation. It produces exactly the kind of uncertainty that makes companies pull back, investors hesitate, and talent look elsewhere.
The people are watching this. Not just the AI industry. Not just the cybersecurity community. Regular people who are trying to figure out who they trust with decisions this consequential. People who don’t follow AI policy closely but understand instinctively when something looks unfair. When a company gets 90 minutes and no explanation, that lands differently than a formal regulatory action with published criteria and a comment period.
It lands like a warning.
And warnings without clear standards don’t produce compliance. They produce fear, resentment, and eventually resistance. The public will form a view about AI governance whether the government builds a framework or not. The question is whether that view gets formed around a process people can see and evaluate — or around incidents like this one, where the reach is visible and the reasoning is not.
The Faust Baseline exists because this gap was always coming. Not because anyone predicted this specific incident. Because the structural condition was obvious from the beginning. AI capability accelerates faster than governance frameworks can form. In that vacuum, whoever acts first sets the terms. And acting first without a transparent standard isn’t leadership. It’s just speed.
Ninety minutes is not governance.
It is a demonstration of what happens when there are no terms at all. The people will decide who gets to set them. That decision is closer than anyone in Washington seems to understand.
Contact: micvicfaust@gmail.com
Post Library – Intelligent People Assume Nothing
The Faust Baseline™ — intelligent-people.org
Codex 3.5 | Twenty Protocols | Ratified and dated on the public record.
Purchasing Page – Intelligent People Assume Nothing
© 2026 The Faust Baseline LLC | All Rights Reserved
