Why independent oversight of AI can no longer be delayed

When was the last time a technology capable of reshaping the foundations of civilization developed practically without state oversight? Suzanne Nossel, head of PEN America and member of Meta's independent oversight board, believes this is precisely what is happening with artificial intelligence — and that the window for action is rapidly closing.

In her column for The Guardian, Nossel draws an instructive comparison. Radio, nuclear energy, the internet — each of these technologies developed from the outset with active government participation. For medicines, there is the FDA, which tests drugs before they reach the market. For the nuclear industry, there are strict requirements for disclosing accidents and incidents. For artificial intelligence, nothing of the sort exists. Companies release models whose operation is not fully understood even by their creators, and they bear virtually no responsibility for the consequences. Chatbots are already giving teenagers advice about suicide, and experts warn that soon they will be able to instruct on the creation of biological weapons.

Why is regulation stalling? Nossel points to three key factors. First, the colossal lobbying resources of technology corporations, which in Washington rivals only the influence of the oil and pharmaceutical industries. Second, political polarization in Congress, where any AI legislation risks becoming hostage to partisan disagreements. Third, the very nature of the technology — it develops so rapidly that lawmakers physically cannot keep up with understanding what exactly they are trying to regulate. Donald Trump, meanwhile, is attempting to nullify regional AI laws that individual states have begun to adopt in the absence of federal standards. Europe, seemingly ahead with its AI Act, is already facing pressure from business claiming that strict rules undermine the continent's competitiveness.

Particularly notable is Nossel's criticism of companies that position themselves as responsible actors. OpenAI, Google, Anthropic — all proclaim commitment to safety. However, the race to dominate the AI market requires pouring billions of dollars into developing ever more powerful models, adding advertising, and expanding collaboration with military structures. Anthropic, which builds its brand on the idea of the "most conscientious" AI company, uses a thought experiment as an ethical guide — the model should imagine how a "thoughtful senior Anthropic employee" would act. Nossel justly notes that this approach reproduces the very problem Silicon Valley has been criticized for years: a narrow circle of people from a privileged background making decisions that affect billions of users worldwide.

The numbers confirm the scale of public concern. According to 2025 polls, 77 percent of Americans believe that AI could pose a threat to humanity. This is not a marginal position of alarmists — it is an overwhelming majority of the population of the country that is the world leader in developing this technology. Yet between awareness of risks and actual action lies a chasm.

For Russian audiences, this discussion is of particular significance. In Russia, AI regulation is at an even earlier stage than in the US or Europe, and the country's largest technology companies are actively developing their own language models and AI services. International experience — and mistakes — in building oversight systems will inevitably influence the formation of Russian approaches to this problem.

Nossel's position boils down to a simple but important idea: if states are unable or unwilling to regulate AI right now, the minimum step should be companies' agreement to independent external oversight. Not internal ethics committees, not thought experiments about how an ideal employee would act, but full-fledged audits by structures independent of these companies' profits. This will not solve the problem entirely, but at least it will create a first line of defense — until lawmakers catch up with the technology that has long since overtaken them.