Warner on AI regulation: ‘We probably can’t solve it all at once’
The Virginia Democrat has seen lawmakers repeatedly fail to pass new tech regulation despite rising concern about issues like privacy and misinformation. He says a different approach is needed for artificial intelligence.
Sen. Mark Warner (D-Va.) and his colleagues have taken big swings at tech legislation in recent years only to come up short, so he’s urging a different approach as Congress looks to regulate artificial intelligence. His pitch: be less ambitious.
“Everybody up here on the Hill acknowledges the downside of the notion that we’re going to simply have the tech guys do the same thing. ‘Well, let us figure it out first and figure out the rules later’ is, I think, a bad proposition,” Warner said in an interview on the POLITICO Tech podcast.
“I’m very sensitive to the notion that on AI we shouldn’t do that,” he continued, “but if we try to overreach, we may come up with goose eggs” — meaning nothing.
Congress sat on the regulatory sidelines throughout the rise of the internet and social media, only to later discover widespread concerns including data privacy, hate speech, election interference, misinformation and market dominance. Even after years of tense hearings and legislative proposals, Warner acknowledges “our record in social media is a big fat zippo.”
He worries lawmakers will suffer a similar fate with artificial intelligence by trying to mitigate its full spectrum of risks with a single law — comprehensive legislation that others, including Sen. Todd Young (R-Ind.) also doubt is realistic. Instead, Warner’s been selling his colleagues on first tackling narrowly focused issues: the potential for AI-generated deepfakes to disrupt elections and financial markets.
“Where I’m at on the regulatory front is we probably can’t solve it all at once,” Warner said. “But where are the two most immediate areas where AI could have an almost existential threat tomorrow?”
He said he’s considering new regulations that would perhaps address concerns about bias or require labels for AI-generated deepfakes — though Warner said he has reservations about allowing companies to apply labels.
He’s also weighing an increase in penalties under existing laws when AI is used to undermine elections or markets. But who might pay those penalties when technology is abused — the tech company or their users — has been a sore point for Congress. A law created at the dawn of the internet, known as Section 230, has largely shielded tech companies from liability for their user’s actions.
“Even the biggest advocates of Section 230, in my conversations with them up here on the Hill, have said they don’t expect Section 230 to carry over to AI,” Warner said.
A targeted bill would still struggle to clear a sharply divided Congress, especially one that deals with election security, Warner said. But he argues it stands a better chance than some of the more sweeping ideas being considered, including the notion of creating a federal agency to oversee AI. Warner said he’s not against that idea, but with a Republican-controlled House, “I wouldn’t put all my eggs in that basket.”
Attempts to regulate technology with ties to China, in particular the video-sharing app TikTok, offer another cautionary tale, Warner said. Legislation that Warner introduced earlier this year with Senate Minority Whip John Thune (R-S.D.) that would give the Commerce Department more oversight of foreign-owned tech firms, called the RESTRICT Act, S. 686 (118), was “lining up senators two by two, like Noah’s Ark” and had the White House’s blessing before stalling amid political attacks earlier this year.
Warner said he has less anxiety today about China dominating AI than he did a year ago, though concerns remain about Beijing using the technology to advance its military and intelligence operations. But if Congress cannot come to a bipartisan agreement on how to combat national security concerns posed by Chinese technology, he said, then the prospect for comprehensive AI legislation looks grim.
“It’s so important, this is more on the politics side than the substance side, to at least show we can do something now,” Warner said. Even if industry and other groups think that’s all Congress will do, he added, “I will take that risk because we’ve been so pathetic on social media. We’ve got to show that we can actually put some markers down that have the force of law.”
Annie Rees contributed to this report.
For a daily download on tech, politics and policy, subscribe to POLITICO Tech on Apple, Spotify, Google or wherever you get your podcasts.