That the leaders of the most powerful artificial intelligence companies are all calling for regulation shouldn’t be ignored. But big businesses calling on big government to cement their status as industry leaders isn’t new. And lawmakers should exercise extreme caution when attempting to regulate an industry they don’t understand, especially one in which so many past warnings have turned out to be false.
Anthropic CEO Dario Amodei called for slowing the development of frontier AI and bringing outside evaluators into the process this past weekend. OpenAI CEO Sam Altman, Google DeepMind CEO Demis Hassabis, and SpaceXAI CEO Elon Musk all endorsed the idea.
Former Obama White House chief of staff Rahm Emanuel then opined that this unanimity should terrify us. “I can tell you when the last time a CEO or an industry at large asked to be regulated: never,” Emanuel posted on X, arguing that the requests for oversight prove the industry itself knows it is careening toward danger.
But Emanuel’s premise is simply false.
Powerful businesses ask the government to regulate their industries all the time, particularly when the regulations being discussed would help powerful businesses.
Before FTX collapsed, Sam Bankman-Fried became one of Washington’s most enthusiastic cryptocurrency regulation advocates. He pushed legislation that would have placed much of the industry under the Commodity Futures Trading Commission, an agency widely regarded by crypto firms as friendlier than the Securities and Exchange Commission. Regulation would have given crypto greater legitimacy while creating rules that a large, centralized exchange such as FTX would be well-positioned to navigate.
Mark Zuckerberg did something similar after Facebook endured years of controversy over privacy, political advertising, and content moderation. In 2019, he explicitly called for the government to play “a more active role” in regulating the internet, including rules covering harmful content, election integrity, privacy, and data portability. Such legislation would have replaced some of the legal uncertainty surrounding Facebook with government-written standards that a company of Meta’s size was well equipped to meet, while potentially limiting the company’s exposure to future lawsuits.
Health insurers provide an even clearer example. During the debate that produced the Affordable Care Act, the insurance industry agreed to accept customers with preexisting conditions in exchange for an individual mandate requiring Americans to buy insurance. Aetna CEO Ron Williams was among the industry leaders promoting reform. The eventual law did not merely regulate insurers; it required millions of Americans to become their customers and subsidized coverage for many of them with taxpayer dollars.
Businesses do not ask Washington to regulate them out of altruism. They ask for rules they believe will benefit them.
That should make Congress particularly cautious about rushing to regulate AI based on predictions from the executives developing it. Altman himself has acknowledged that several of his forecasts about the speed of AI disruption were wrong.
He expected the release of GPT-4, the technology then powering OpenAI’s ChatGPT, to make software businesses “up for grabs right away.” It didn’t. He expected far more entry-level white-collar jobs to have disappeared by now, later saying he was “delighted to be wrong.” And at the beginning of 2025, he predicted AI agents would “join the workforce” that year and “materially change the output of companies.” That revolution did not arrive on schedule.
None of this means AI poses no dangers. Powerful systems can be used irresponsibly, and companies should not be able to escape responsibility merely because an algorithm performed the harmful act. Congress should examine how existing criminal, civil, and product liability laws apply to AI and make sure the people and companies making consequential decisions can be held accountable when those decisions cause foreseeable harm.
EDITORIAL: AGAINST TRUMP’S $5,000 BRIBE
But accountability is very different from creating a new federal bureaucracy empowered to approve models, license developers, or bless every new technological development before Americans may use it.
The history of regulation is filled with incumbents asking the government to write rules that incumbents are best equipped to survive. AI may someday require new laws. But the fact that today’s dominant AI companies are asking for them is a reason for Congress to scrutinize those proposals more carefully, not rush to enact them.