Who will regulate artificial intelligence?

President Trump's meeting with American tech giants this week further clarified the camps in the debate over who will set the rules for the technological development that will transform the global economy, national security, and technological progress. Trump's signing of an agreement based on industry self-regulation, in order not to hold back the AI industry, stands in complete contrast to the approach of politicians like Bernie Sanders, who prioritize public oversight. The memorandum signed on September 29, which the Trump administration describes as "morally binding," envisions that AI companies will be subject to internal oversight and external oversight by an independent board, but the agreement has no sanctioning power. Trump does not want this industry, which he wants to rebrand as "superintelligence," to be preoccupied with rules that would make it lose the race with China. America's debate over how AI will be regulated, at least for now, is proceeding in the direction the tech giants want.

THE LOGIC OF THE AGREEMENT

The fact that the AI safety debate, which reached its peak in recent weeks, was brought to the agenda by industry leaders indicated that the matter was not solely about safety. Both the future of the trillions of dollars of investment made in this sector and the determination of the rules of the race with China stood out as critical issues. The agreement envisions that tech companies establish their own internal security mechanisms, apply for independent audits, and have overseers on their boards in order to reduce the risks that increasingly powerful AI systems may create. The fact that this arrangement is based on voluntariness and has no federal binding force will prevent companies from being bogged down by overly restrictive rules while slowing down their pace of AI development. Of course, this is bad news for those who argue that these companies have become excessively powerful and that it is impossible for them to oversee themselves.

Those on the side of the industry have reasonable explanations, because it is impossible for legislative processes to keep up with the speed of AI technologies. The companies developing the most advanced models possess a great deal of technical knowledge and competence that the federal bureaucracy does not have expertise in. Therefore, heavy rules and rigid regulations carry the risk of both being unenforceable and slowing down the sector. Moreover, it is impossible not to partially acknowledge those who argue that this would give China a strategic advantage. However, it is not realistic to expect that companies' self-regulation mechanisms will prioritize the public interest over commercial purposes. The companies investing hundreds of billions of dollars to develop these systems are under pressure both to maximize their commercial returns as soon as possible and to win the race to possess the most advanced technologies.

TECHNOLOGY OLIGARCHS

Bernie Sanders's criticism, too, focuses not on the voluntary nature of the agreement but on the concentration of immense economic power in the hands of a certain elite stratum of technology and this group's political influence. It does not escape attention that these names, whom Sanders describes as "oligarchs who run the country," spend hundreds of millions of dollars lobbying Congress. In Sanders's words, "Congress is not regulating AI; AI is regulating Congress." The influence of tech giants over Trump in particular and over Congress is undeniable, because these companies, which have served as the engine of the American economy in recent years, also contribute to employment, albeit in a limited way, by building data centers. Although many local politicians complain that data centers raise energy and housing prices, they also do not want employment to flee to other states or districts.

In March, Trump, addressing the data center problem, supported the initiative in which Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI promised not to pass their electricity costs onto the public. The companies promised to produce, procure, or generate the energy their data centers need themselves, and not to reflect the necessary infrastructure costs onto ordinary consumers' bills. But this political debate has not disappeared, and grid costs have also begun to be debated in Congress. This shows that voluntary commitments have not been enough to dispel public concerns. The transformation of AI's oversight and governance problems into electoral politics in this way creates a risk for Republicans in the November elections, especially for those who want to keep the sector's path open. Sanders's pursuit of this issue indicates that he wants to brand the Trump administration as one that has surrendered the country to technology oligarchs and is shifting the bill for AI development onto citizens.

CHINA'S SHADOW

All AI debates circle back to the fierce competition with China in advanced technologies. If the sector is regulated and overseen with heavy rules, slowing American companies and allowing China to close the gap, Washington fears losing leadership in an area that will be decisive economically and militarily. The claim of sustaining rapid innovation while keeping it under control through companies' own security mechanisms can be called a guarantee that security will be pushed to the back burner. This race, in which the US and China define falling behind each other as strategic defeat, may also increase the risk tolerance of companies and states. It is obvious that America needs both technological progress and regulation, but in this race under China's shadow, the answer to the question of whether final authority should rest with the private sector or the state has not been given.

The memorandum Trump signed with the tech giants appears to have postponed the debate over AI oversight rather than resolving it. An agreement based on voluntariness and without sanctioning power ensures that the industry does not slow down in the race with China, but it does not create a binding mechanism. The fact that Sanders's emphasis on oligarchy is finding resonance in an election year, that the costs created by data centers are being carried into local politics, and that pressure from Congress is increasing could play a game-changing role in the November elections. Companies' self-regulation looks like a model that will hold up until a major scandal occurs, but it is clear that it is not a permanent solution. In the coming period, Washington's real test will be whether it can establish a regulatory framework capable of overseeing AI without losing the race with China.