Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Tim Wu’s answer, in a May 2023 interview, was that Washington was looking at AI too narrowly: it needed to address concrete consumer harms and the power of dominant firms without making compliance so costly that smaller competitors could not enter. He favored enforcing existing consumer-protection laws, requiring AI systems to identify themselves, examining gaps in laws written around human conduct, compensating creators, and investing in open-source alternatives. These were Wu’s proposals—not enacted policy or proof that any particular regulatory approach would produce the outcomes he feared or wanted.
What did Wu think Washington was missing?
Wu, a former White House adviser associated with President Biden’s antitrust policy, had left the administration in January 2023 and returned to Columbia Law School. The Washington Post reported on May 30, 2023, that he was meeting with officials to discuss AI regulation.
His central concern was a mismatch between the debate and the problems he saw emerging. Attention to hypothetical or abstract risks, he argued, could distract from existing harms such as deceptive AI-generated product reviews and scams. At the same time, he worried that rules designed with the largest technology companies in mind could entrench their position by imposing costs that smaller firms would struggle to bear.
That is a policy argument about trade-offs, not an empirical finding that regulation necessarily helps incumbents. Wu’s question was how to protect people while keeping the market open to challengers and new ideas.
#1 Best Overall
Why did Wu oppose licensing AI companies?
Wu opposed a licensing system for operators of large AI models because he believed it could make regulatory compliance a barrier to entry. In the interview, he put the concern bluntly: “Licensing regimes are the death of competition in most places they operate.” He also said he opposed an approach that created “heavy compliance costs for market entry” while regulating more abstract harms.
Those statements express Wu’s judgment; the interview did not quantify the likely costs or establish that a licensing regime would have the same effects in every market. The underlying design question is whether requirements can be met by smaller entrants or whether they effectively reserve participation for firms with substantial legal, technical, and financial resources.
Should AI systems have to identify themselves?
Yes, in Wu’s view. He argued that people should be told proactively when they are interacting with an AI system, rather than having to ask a chatbot whether it is AI. He suggested that an agency such as the Federal Trade Commission (FTC) could develop a format for complying with such a requirement.
Wu presented identification as a practical consumer-protection measure. If a person does not know that a review, message, or conversation came from AI, it may be harder to assess its reliability or recognize deception. The Washington Post interview cited misleading AI-generated product reviews as an example of the kind of concern disclosure could help address.
Why not create a new federal AI agency?
Wu was wary of establishing a new AI-focused regulator. He worried that a new agency, particularly one with costly or broad requirements, could advantage companies already able to meet them and freeze the industry before it got started. He preferred using existing institutions and laws for problems they can already address.
This was not an argument that every AI-related problem already has a clear regulatory answer. Wu paired his preference for existing enforcement with a call to identify legal gaps. The distinction is between applying established rules to conduct they cover and creating new authority where current law does not provide a clear remedy.
How did Wu think existing law could address AI harms?
Use deception and fraud rules where they fit
Wu argued that regulators could apply existing prohibitions on deceptive and misleading practices to AI abuses. A requirement that systems identify themselves could help people recognize some scams or misleading interactions, while enforcement could target conduct that already falls within consumer-protection law.
Disclosure would not solve every problem. Wu supported transparency but cautioned that it was insufficient on its own: “It’s not bad, but it’s not enough.” Knowing that AI produced something does not, by itself, prevent harm or establish who is legally responsible for it.
Look for gaps where laws assume a human actor
Wu also questioned whether laws built around human intent, malice, or recklessness map neatly onto harms caused by AI systems. He called for identifying situations in which AI could cause harm but existing law offered no clear remedy. In his words, “We have a pressing need to figure out the areas of the legal code that are likely to be violated by an AI likely to cause harm, but where the laws are written with a human in mind.”
He suggested that the Justice Department could identify such areas and Congress could address gaps through legislation. The interview referred to this project as a “robot penal code”; that phrase described a proposed examination of legal coverage, not a law that had been enacted.
What did Wu propose for creators whose work trains AI?
Wu proposed considering a licensing mechanism modeled on payments to composers when music is played on the radio: creators whose work trained AI models could receive proportional compensation. The interview noted that the mechanism was still under debate. It did not establish a settled system for identifying training material, setting rates, distributing payments, or resolving disputes.
The proposal highlights a difficult design balance. A workable licensing approach would need to address how payment is calculated and allocated without making access to training material impractical or concentrating compliance burdens among only the largest developers. Wu raised compensation as an issue to solve, not as a finished policy blueprint.
Why did Wu favor public investment and open-source AI?
Wu supported public funding for technology research but warned against using subsidies to accelerate AI expansion by large, already profitable companies. He also argued for public investment in open-source models as a way to encourage wider innovation and counter concentration. He floated an AI “public option,” drawing an analogy to ARPANET, an early publicly funded network.
The proposal reflects his broader concern that leaving development entirely to private firms could let a small number of companies shape the field. Wu summarized that view in the interview: “For some reason, the last 20 years we’ve assumed everything can happen completely privately, and I think we should learn the lesson from that.” The analogy is a rationale for public involvement, not evidence that a particular public model would succeed or that it should replace private development.
How do Wu’s proposals fit with later federal actions?
Wu’s 2023 recommendations should not be confused with later government actions. A December 11, 2025 White House executive order called for a national policy framework, created a process for challenging certain state AI laws, and directed agencies to assess possible grant conditions. It also instructed the FTC and Federal Communications Commission (FCC) to take specified steps concerning federal disclosure standards and application of the FTC Act to AI models. Those directions came after the interview and do not establish that Wu’s proposals became policy.
On July 1, 2026, the FTC sought public comment on a proposed policy statement addressing AI accuracy and deception, with a July 31 comment deadline. The available account describes a proposal, not a final rule or policy; its status after the deadline is not established here.
Free tools Windows power users keep installed
One-click scans. No signup required.
A September 29, 2026 White House executive order instructed executive agencies to use “Super Intelligence” and “SI” instead of “Artificial Intelligence” and “AI” in specified non-statutory communications. It retained the existing statutory definition for implementation unless superseded by further action. This is a terminology instruction for executive-branch communications, not evidence that the underlying technology or statutory framework changed.
What is the practical takeaway?
Wu’s framework is a balancing act: target concrete deception with tools already available, require clear identification, and investigate where law fails to cover AI-caused harm—while resisting rules or subsidies that, in his view, could reinforce the biggest firms’ advantages. His interview offers arguments and policy options, not measured predictions. Its enduring question is how to protect consumers and creators without making the ability to comply a privilege of incumbents.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
