The gap between the promise of AI regulation and the political reality of regulatory capture has never been wider.
Dario Amodei, CEO of Anthropic, recently released a letter titled “We Must Pace the Frontier.” On the surface this is a reasonable document because it calls for measured, thoughtful regulation of the most powerful AI systems humanity has ever built. In theory, I agree. Frontier models pose genuine risks like catastrophic misuse, concentration of power, and the erosion of human agency. A sensible regulatory framework could, in principle, channel this technology toward human flourishing. Yet here is the uncomfortable question that the letter does not answer: Who writes the rules and what do they prioritize?
“There is nothing inevitable about a future where artificial intelligence becomes another instrument of warfare, surveillance or extraction.” -Emilio Miller-Lopex, Founder of GreenSeek.AI
Autonomous Killing, Warfare, and AI Blunders
AI targeted drone bombings and killer robots used to only be the stuff of dystopian films like Terminator. On January 12, 2026 Pete Hegseth’s Department of War announced a “transformative AI Acceleration Strategy to Secure American Military AI Dominance”. A September 18 report by CNN details how an AI report falsely identified a Chinese ship in the Middle East as transporting nuclear material. This happened last spring and nearly could have sparked World War 3 yet politicians on both sides of the aisle were strangely silent on the issue. Yet somehow these same lawmakers were eager to call for AI regulations based on tweets from Darious Amodei, Jacob Coxon and Elon Musk. Is Congress so out of touch that they only respond to billionaire tweets while ignoring real-world events like the one described in the CNN article?
“We will become an ‘AI-first’ warfighting force across all domains.” -Pete Hegseth, US Department of War
Regulatory Capture is the Norm in the US
In the US we have a long and well-documented history of regulatory capture. This is when the very industries meant to be regulated end up controlling the agencies that oversee them. The pattern is so consistent it has a name in political science, “the revolving door”. Executives cycle between corporate boardrooms and regulatory commissions, carrying with them the assumptions, incentives, and lobbying budgets of the firms they are supposed to police.
When Amodei calls for “pacing the Frontier” we must ask some very important questions. Will the resulting regulations be used to protect the public, or will they will be written in a way that protects the monopoly of the world’s largest companies? If only the largest, wealthiest labs can comply with these regulations that will lock out smaller, ethical competitors like GreenSeek and others.
This sort of monopoly would not only be a monopoly on AI technology. It would effectually be a monopoly on the distribution of information. This would allow a few companies to decide what can and can’t be discussed, along with who will get censored. As you can imagine, this is a very slippery slope.
The Promise: What Amodei Actually Says
Amodei’s letter is earnest. He argues that AI development is moving faster than our institutions can adapt, and that we need deliberate, forward-looking governance rather than reactive panic. Yet the media, experts, activists, and certain political leaders have done nothing but focus on panic with endless doomsday scenarios about human extinction. Amodei frames regulation not as a brake on innovation but as a way to keep the field safe and sustainable over the long run.
He is not the first person to bring attention the stakes of this emerging technology. Frontier models are being deployed in contexts that touch everything from healthcare to warfare. His suggestion of a “pause” or a “pace” is not inherently anti-innovation. It is merely a recognition that some technologies are too consequential to be left entirely to market forces.
There is theory and there is the political reality in practice. This is where the two collide.
The largest frontier labs do not fear regulation. They fear competition.
The Reality: Examples of Regulatory Capture
Regulatory capture is not a conspiracy theory. It is a documented institutional pattern. Consider the U.S. Food and Drug Administration (FDA). In recent years, a significant portion of the agency’s budget has come from user fees paid by the very pharmaceutical companies it regulates. The result is a system where drug approvals are often expedited to accommodate industry timelines. Critics argue the agency has become more responsive to its funders than to patients and most informed voters would probably agree.
The Federal Communications Commission (FCC) offers another example. The agency is nominally tasked with protecting consumers and ensuring fair competition in telecommunications. Yet it has repeatedly been led by former industry executives who rolled back net neutrality protections. These are the rules that smaller competitors and consumer advocates had fought for years to establish.
This is the “capture cycle” where industry pushes for deregulation when friendly officials are in power, and then pushes for “compliance-heavy” regulation when it wants to raise barriers to entry against smaller rivals.
After the 2008 crisis, the Dodd-Frank Act was supposed to rein in the largest banks. But a decade of lobbying and litigation hollowed out many of its strongest provisions. This allowed the “too big to fail” institutions to emerge from the crisis larger than before. The regulators had been captured because the industry could outspend, out-litigate, and out-staff any public-interest group.
We must view Amodei’s call for pacing the frontier through this lens!
The Bait and Switch: Regulation as a Moat
Compliance is expensive. Safety evaluations, red-team testing, audit trails, and government reporting all cost money. Meanwhile the cost scales far more favorably for a company with billions in revenue than for a small lab running open-weight models on renewable energy.
The structural incentives are clear. A regulatory framework that requires extensive compute reporting, mandatory safety certifications, and liability insurance creates a moat around the incumbents. Smaller players using open-source, open-weight labs that cannot afford armies of compliance lawyers will find themselves priced out of the market.
This is the classic “bait and switch” of industry-friendly regulation. In theory, the rules protect the public. In practice, they protect the largest firms from disruption.
Regulations Do Not Replace Basic Ethics
Amodei’s letter, and the broader discourse around AI safety, focuses almost exclusively on catastrophic risks like biosecurity, cyberattacks, and loss of control. These are legitimate concerns. Yet the regulatory conversation is strikingly silent on ethical issues like autonomous weapons, where AI has been used to target innocent civilians in war zones. Surveillance, copyrights, data sovereignty, privacy, and censorship and other basic ethics are suspiciously missing from the debate.
“We have made a commitment not to participate in the militarization of artificial intelligence, not to build technology for autonomous weapons, and not to collaborate with mass data-surveillance systems.” -Emilio Miller-Lopez, Founder of GreenSeek.AI
Legal Precedent to Hold Companies Liable Now
Historically when large companies create a dangerous or faulty product they are held accountable in numerous ways. Lawsuits and federal indictments are already an option for rogue AI agents that break applicable laws. There is no need to add more regulations when these companies should be prosecuted now.
In the 1970’s when the Ford Pinto had a design flaw that caused them to catch fire in a collision, they were held responsible. There were over 100 civil lawsuits where the company was forced to pay out tens of millions in settlements. Most of all Fords reputation was decimated by the market. There are numerous other examples that can and should be applied in the current rogue AI criminality.
This brings us another angle on the current situation with large AI companies and their Frontier Models. The public has a say because they vote with their dollar. When Anthropic announced that it didn’t want to participate in autonomous killings with their technology, many people dropped Chat GPT and brought their business to Claude. If there is transparency, regulations are not needed because the consumer will choose products that align with their values.
Let us Not be Fooled by the Billionaires Profiting on Frontier Models
The industry is betting that we will not notice the difference between the regulation promise and the political reality. The children currently being targeted and bombed in war-zones are already paying the price for these unethical companies products.
Will we choose AI for regeneration, to uplift humanity, and for peace? Or will we allow greed and unethical companies to apply this technology to oppress the general public, censor independent voices, target civilians in a war-zone, and hack government databases all while consolidating wealth and power into the hands of a few? The choice is ours. Together we get to shape the future of AI. There is a better way.
