In a Breitbart op-ed published Thursday, Foundation for Freedom Online managing director Allum Bokhari argues that the sudden, coordinated push to regulate artificial intelligence follows a familiar political script — one he says has repeatedly been used to expand government power, only to leave the public with institutions that outlive their original justification.
Bokhari opens by asking what the Federal Reserve, the Patriot Act, Net Zero policies, COVID lockdowns, foreign disinformation programs and the proposed globally coordinated AI slowdown all have in common. His answer: each was preceded by a crisis, “real or imagined,” that planted existential fear in the public and created the political cover for a sweeping government response — a response that was often met with initial relief before its costs became clear.
The pattern, as Bokhari lays it out
The op-ed traces a chain of examples. FISA courts expanded to fight Islamic terrorism, Bokhari writes, were later turned toward surveillance of a domestic presidential campaign. COVID lockdowns persisted in some places after jurisdictions such as Sweden, Florida and Texas demonstrated they were unnecessary, in his account. Net Zero policies, he argues, have left Europe economically weakened and dependent on foreign energy.
In each case, according to Bokhari, the result was a broader expansion of the federal bureaucracy — new departments, new titles and responsibilities, and a sprawling network of NGOs, researchers and “civil society” groups positioned to draw on newly opened streams of taxpayer money tied to the declared emergency.
A case in point he cites: the short-lived “Director of the Disinformation Governance Board” at the Department of Homeland Security, a post created during the Biden administration and abandoned after intense criticism.
Why AI is the new frontier
Bokhari points to AI regulation as the current frontier. Within roughly two weeks, he notes, the two leading AI companies — OpenAI and Anthropic — both asked the government to step in to slow the industry down collectively. That followed the resignation of an Anthropic researcher who said superintelligent AI could “kill us all,” a claim Bokhari says drew more than a hundred million views on social media and was carried on primetime broadcasts worldwide.
What came next, in his telling, was the rapid activation of a global governance apparatus: 70 UK parliamentarians calling for a ban on superintelligent AI, former President Barack Obama and former Vice President Kamala Harris joining the conversation, and another British politician urging the United Nations to intervene.
Bokhari also flags legislation introduced just days before the Anthropic resignation by Sen. Bernie Sanders (I-VT) that would ban AI superintelligence and establish a new federal agency dedicated to AI safety. Under the bill, as he describes it, anyone who developed frontier AI models without government approval would face up to twenty years in prison.
He reserves particular suspicion for Netflix’s decision, two days before the op-ed’s publication, to drop an AI apocalypse documentary featuring interviews with several of the figures calling for government intervention — among them Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman and Eliezer Yudkowsky. Because such documentaries take considerable time to produce, Bokhari writes, the timing has fed allegations of coordination.
Bokhari then returns to his central comparison: the panic of 1907 that produced the Federal Reserve, 9/11 that produced the Department of Homeland Security and an expanded FISA regime, and the “largely imagined” Russian disinformation panic of 2017 that produced a vast government-funded counter-disinformation apparatus — one he says was eventually used to censor Americans, including the sitting president and the sitting HHS secretary at the time of writing.
The ’embedded evaluators’ problem
The op-ed zeroes in on one specific element of Amodei’s proposal to “pace the frontier”: a call for all AI companies to grant access to “embedded evaluators who have employee-like access to verify safety practices and report incidents.”
Bokhari says the country has seen this arrangement before. It began, he argues, with external “trust and safety” councils given access to content moderation systems at platforms like Twitter, whose members included LGBT and feminist activist groups. It ended, he writes, with a globe-spanning network of trust-and-safety, brand-safety and counter-disinformation organizations — often funded by U.S. government agencies — that contributed to mass censorship of social media.
He warns the “AI evaluators” Amodei wants embedded in every company are likely drawn from the same milieu. As evidence, he cites Model Evaluation and Threat Research (METR), a nonprofit he identifies as Amodei’s preferred watchdog. Bokhari points to a presentation by METR’s CEO, standing before a slide featuring migrants on a raft, in which the executive reportedly said future generations would be “embarrassed” that nations barred people from a location based on “where they happen to be born — their nationality.” Bokhari characterizes that position as being embarrassed by the existence of national borders.
He also questions METR’s funding. The nonprofit raised $71 million in just six months, according to Bokhari, but researchers tracing the money have accounted for only about $350,000. The rest, he writes, remains dark.
Bokhari argues that if AI safety advocates genuinely believe humanity’s future is at stake and want a bipartisan consensus, they have an obligation to explain how their proposed measures would not hand control over the most consequential technology of modern times to people seeking to impose what he calls fringe left-wing viewpoints on everyone else.
Demanding specifics instead of vagueness
The op-ed takes aim at the vagueness of AI doom predictions, which typically assert that a superintelligent AI would be more intelligent than humans without specifying what it would actually do. Bokhari concedes there are strong arguments that large language models like ChatGPT or Claude will never be meaningfully superintelligent, but says the debate should proceed on the doomers’ own assumptions.
If AI is correct, he asks, how would a superintelligent system attack humanity? The most plausible scenario, in his view, is AI agents seizing a biolab and manufacturing deadly pathogens. If that is the central fear, he argues, then it is the limiting principle — and there is no justification for empowering regulators whose concerns extend to unrelated matters, such as national borders.
Once the specific threat is defined, Bokhari continues, the question of how to address it becomes tractable. Securing nuclear facilities against cyberattacks? The International Atomic Energy Agency and the Pentagon have worked on that for decades. Securing biolabs? He lists the International Experts Group of Biosafety and Biosecurity Regulators, the International Biosecurity and Biosafety Initiative for Science, and the Association for Biosafety and Biosecurity as bodies that already exist.
“How about that,” he writes, the global regulatory bodies for the two areas of greatest concern already exist. In his view, the case for handing control of the technological future to “open borders supporters and Effective Altruist weirdos” collapses once specifics are demanded — which, he suggests, is precisely why the doomers avoid them.
“Pace the frontier,” Bokhari concludes, is only possible if the public insists on specifics rather than accepting the vagueness that, in his account, justifies government expansion without end.
Allum Bokhari is managing director at the Foundation for Freedom Online, a digital censorship watchdog. The op-ed was published by Breitbart.
Source: www.breitbart.com — https://www.breitbart.com/tech/2026/09/18/bokhari-we-must-pace-the-frontier-of-government-power-grabs/