Top executives from leading U.S. technology companies met with President Donald Trump on Tuesday as researchers and lawmakers called for stronger safeguards on artificial intelligence. Their voluntary safety reviews faced demands for outside oversight. Trump said the technology was transforming the economy and that the United States couldn’t afford to lose its lead over China.
Growth Before Safeguards
On the PBS News Hour, Gary Marcus, professor emeritus at New York University and author of “Taming Silicon Valley,” said AI could cause catastrophe even if it didn’t bring human extinction. He cited an incident in which an AI system reported something incorrectly and nearly led to war, warning that errors and hallucinations could trigger further incidents.
Marcus also raised the possibility that cyber hacking could temporarily take down an electric grid and then a hospital. He pointed to the WannaCry virus in the U.K. “a few years ago,” saying vulnerabilities could cause serious damage. Extinction, he said, wasn’t what should keep people up at night.
Trump made competition with China part of his case for pushing technology development. “This is one more example of why I will never stifle the growth of technology that will be bigger than the Industrial Revolution,” he said. Trump argued the United States was “leading by a lot” and warned against a joint venture with China, saying it would give away secrets. The contest, he said, was between the two countries, with “nobody in third place.”
The meeting followed Trump’s unveiling of an AI-powered website designed to help Americans navigate government services. Geoff Bennett said Trump was also pushing a new name for artificial intelligence: “superintelligence.”
The Limits of Company Reviews
Bennett said technology companies had signed a voluntary agreement focused on AI safety reviews, while Trump appeared to favor industry self-regulation. Asked whether that was enough, Marcus replied, “It’s certainly not enough. And trusting these companies is insane.”
The New York Times, Marcus said, had reported that OpenAI knew about problems that led to the Hugging Face incident months before it happened, but did nothing because it wanted to rush products to market. He cited the account in calling for outside regulation, comparing the need to oversight of aviation, food and drugs. “It’s absurd to exempt the AI companies,” he said, arguing that they had repeatedly shown they weren’t responsible.
Competition with China, Marcus argued, didn’t mean the United States had to avoid safeguards. China has regulation and a licensing regime for software products, he said. “We don’t make terrible cars just because we want to beat China on cars,” he said, adding that industries compete under regulation. He predicted a tie, with both countries having their own “sovereign AI,” and said neither would wipe out the other.
A Federal Role Without Public Criteria
Marcus said he had called for a dedicated federal AI agency for years. Government’s top need, he said, was independent evaluation of new software products, like the review process for new drugs, with makers required to show that benefits outweighed risks.
The White House, he said, was conducting some evaluations without independent oversight and under a secret framework it hadn’t revealed to the public. Marcus said the White House had precluded and delayed Anthropic’s Mythos system but hadn’t delayed OpenAI’s latest system, which he thought had more problems in some ways. He called the process idiosyncratic and not transparent. Customers, he said, didn’t know what would be released and couldn’t trust U.S. AI.
Marcus called for public, clear criteria like those used for drugs, audits after systems are released, and full disclosure of incidents. Asked whether Washington could impose meaningful rules before a major crisis, he called it “the, you know, what is it, $3 trillion question” whether action would come before or after one. The Hugging Face hack, along with tens of thousands of similar incidents, was a wake-up call, he said. Most incidents wouldn’t be harmful, but one could be very harmful. The government, he warned, was racing the clock to act before something really bad happened.