AI doom predictions won’t come to pass. But there’s a bigger threat we should be worried about: regulatory capture. If big AI companies succeed in weaponizing fear to get the government to slow down their competitors, we will all be worse off.
When former OpenAI and Anthropic AI researcher Jacob Coxon claimed that frontier companies were “gambling with our lives” with respect to AI model development and the risks associated with continued growth toward superintelligence, he kicked off a firestorm of online discourse. In quick response, both OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei called for AI development to slow down. Amodei’s three-step proposal directly calls for government standards and enforcement. Altman also suggested a government rule change to suspend antitrust rules may be in order.
Expanding the state is not a good answer to AI doom rhetoric — we have been down this road before. We should be highly skeptical of embracing these calls for regulation. This type of appeal should remind us of disastrous policy decisions from recent history, which have led to business collusion and consolidation at the expense of regular people.
Consider the consequences of the American Medical Association’s panic over what it perceived as an overabundance of doctors in the 1980s. It claimed a surplus would mean some doctors wouldn’t perform certain procedures frequently enough to maintain peak proficiency, while other inactive specialists would shift to general practice without adequate preparation. With the quality of healthcare on the line, the AMA, which once trusted the free market to balance the supply and demand, changed its tune, saying, “Market forces cannot be relied upon by themselves.”
New medical schools were placed on hold. Student enrollment was capped. Even Medicare funding for residences was limited. Over two decades of these policies have now put us on track for a shortage of up to 124,180 physicians by 2027, per one analysis. Anyone who has visited the emergency room or even attempted to schedule a general checkup recently can feel the painful effects of the shortage we now face.
The result of getting the government involved like this is to protect existing professionals and establishments from competition, at the expense of patients and consumers. In the name of patient safety, top voices in the medical industry worked with government regulators to insulate themselves from having to offer better prices or better-quality service.
This is just one of many examples, from U.S. railroads to New York’s taxi medallions. It is this very pattern we see with the frontier AI companies, too. While OpenAI and Anthropic’s market share is undoubtedly large, those companies have yet to achieve pure profitability. This means open-source models and startups, which can leverage lower cost performance, have a chance to catch up.
Is there real potential for AI to be dangerous? Perhaps, but it does not follow that allowing the current power players to dictate policy by way of government intervention is the correct remedy. AI doomers rarely articulate how superintelligence actually leads to humanity’s demise, other than vague statements about security vulnerabilities and potential biomedical weapon development. There is even evidence to suggest some of the “rogue AI” incidents that have stirred this discussion have been exaggerated.
STOP CALLING IT AI. TRUMP’S ‘SUPER INTELLIGENCE’ COLD WAR HAS BEGUN
Frontier models already have incentive to prevent embarrassing public-facing events such as the HuggingFace attack. These companies would be individually and directly liable for damage caused under existing law — perhaps another incentive for them to spread the liability through industry-wide regulations. Similarly, concerns about bioweapons should fall within existing law, and any new regulations we may need should be narrowly tailored to that industry. A blanket approach only helps those already at the top.
Should the concerns of Coxon, Altman, and Amodei worry us? Maybe. But we should be equally worried about the consequences of a manufactured crisis and subsequent regulatory capture. We’re currently experiencing the fallout of rejecting the free market in doctor supply. We should reject similar moves that would cheat us of the benefits of AI.
Donald Kimball is a contributor with Young Voices, focusing on tech policy, free speech, and free markets.