Will regulation let China get ahead, or will it save us from doomsday?
As concerns mount that artificial intelligence has the potential to become a technology humans can’t control and could even wipe out humanity, the debate around what to do — if anything — to slow AI’s development could become a major election issue in the midterms and a legislative battle afterward.
Read more The House is heading home ahead of the midterms even earlier than planned
“AI is taking a lot of power and agency out of people’s lives,” said Noah Giansiracusa, a math professor who studies AI regulation and politics at Bentley University, “so no matter where you are on the political spectrum, you should feel like you are losing control. Because you are.”
Here’s how that debate is shaping up right now.
President Donald Trump and many Republicans have embraced AI as a tool that the United States must develop and own before China does.
“I have concerns that if we don’t win AI, we’re going to be put in a very bad position,” Trump said last week.
The Trump administration has generally been letting Silicon Valley advance AI as quickly as possible, with government getting out of the way. His administration is also investing in ways to develop AI that can deliver talk therapy and dole out medicine and that can supercharge the military. The administration is making some of these decisions in concert with Silicon Valley leaders and Big Tech donors.
“I’d rather they be American killer robots and not Chinese killer robots,” Sen. Ted Cruz (R-Texas) said recently.
There is broad agreement that AI is an incredibly difficult thing to regulate because it touches on so much: health, justice, security.
Some Republicans argue that Congress is too slow-acting to regulate a technology that is developing at warp speed.
“The reflex of legislative bodies is to cover things up with red tape and hyper-regulation,” said House Speaker Mike Johnson (R-Louisiana).
Republican lawmakers generally want to leave it to tech companies entirely to regulate themselves.
Days after a researcher at Anthropic resigned over concerns the company wasn’t doing enough to prevent mass destruction, AI leaders are starting to talk about a handshake agreement to slow development and monitor their own products for safety.
“Where we will need the help of our government is for international coordination,” said Sam Altman, head of ChatGPT maker OpenAI.
Read more In leaked recording, Nebraska Senate candidate criticizes some Trump voters
The fears surrounding AI’s worst-case scenarios are many: It could create political upheaval, lead to the collapse of societal institutions, or create a nuclear or biological threat. Anthropic said it had shut down requests from people trying to use its tools to research diseases such as a mosquito-borne virus because it deemed it possible that those requests could be coming from bad actors.
Reviving language he uses for climate change, Trump has called those fears a “hoax.”
“You’re not all going to be dead in 10 years,” Johnson said recently.
Sen. Bernie Sanders (I-Vermont) has proposed pausing AI’s development until Congress can create an oversight agency. “Stop building machines that humans cannot control,” he has urged AI companies.
This argument resonates across the political spectrum: Sanders spoke this week at a conference calling for a more human approach to AI, as did Trump ally Stephen K. Bannon.
There is a developing consensus on the idea of requiring AI companies to have a “kill switch” to turn all their products off.
“My experience as Intel Chairman taught me the time to stop a crisis is yesterday,” former Republican congressman Mike Rogers, who is running for Senate in Michigan, wrote recently on X. (He’s running in a competitive Senate race in which his Democratic opponent, Abdul El-Sayed, has called for a pause on AI development.)
And Cruz is working on a bipartisan bill that he said would address catastrophic risks involving biological or nuclear threats.
Microsoft’s CEO threw out the idea of embedding independent evaluators at the company and agreeing with other AI companies to deliberately pace development.
Giansiracusa said independent evaluators at companies could help the public better understand what is actually being developed. “The one surefire thing we can do is more transparency,” he said. “The less we know about these companies and how their products work, the more opportunities there are for things to go wrong.”
And whatever you think about doomsday fears, AI is used on a daily basis to commit more mundane crimes, such as fraud, said Laura Edelson, a computer scientist who studies cybersecurity at New York University’s Tandon School of Engineering. “We should ask that the people who make technology have some responsibility to not allow it to be used to bilk the American people out of a billion dollars,” she said.
Amber Phillips writes The 5-Minute Fix newsletter, a quick analysis of the day’s biggest political news. Send her an email here, or ask a question that could be featured in an upcoming newsletter.
Read more Former congressman Sander Levin endorses El-Sayed for U.S. Senate in Michigan
