Artificial intelligence is already sowing chaos and confusion in US elections — from a bogus robocall impersonating President Joe Biden to a falsified hot-mic recording apparently designed to torpedo a Chicago mayoral campaign.
The country urgently needs new laws to prevent deepfakes and other AI-created misinformation from overwhelming elections at an unprecedented scale, policy experts and US lawmakers warn.
But with just nine months until Americans head to the ballot box, there are few signs Congress is ready to pass any meaningful legislation on AI.
Multiple people involved in the legislative process tell CNN their hopes for a wide-ranging AI bill this year are dimming despite a rare, personal push last summer by Senate Majority Leader Chuck Schumer to put AI regulation at the top of the agenda.
After numerous high-profile hearings and closed-door sessions that drew the likes of Bill Gates, Mark Zuckerberg and Elon Musk to Capitol Hill, it appears that typical congressional gridlock may blunt efforts this year to address AI-powered discrimination, copyright infringement, job losses or election and national security threats.
“I’m skeptical that something is going to come together, with the legislative days we have left here, and knowing that there’s bigger priorities for the floor” including must-pass legislation to fund the government, said one congressional aide familiar with the discussions.
Even if Congress does manage to pass a bill regulating AI, it’s likely to be much less ambitious in scope than many of the initial announcements may have suggested, according to a tech industry official, speaking on condition of anonymity to discuss private meetings with congressional offices.
Lawmakers are still publicly expressing optimism about prospects for regulating the AI industry. In a Senate floor speech last month, Schumer reiterated that “Congress must intervene to promote safe AI innovation” and that both political parties “recognize the need to get something done on AI.”
Schumer has previously said that with the election nearing, he may seek to fast-track a bill that focuses specifically on AI and election security. But that could become harder the longer he waits, the congressional aide said.
“Primaries are happening; no one’s going to want to engage,” the aide said.
A spokesperson for Schumer told CNN that he and a group of senators directly leading the charge on AI legislation — New Mexico Democratic Sen. Martin Heinrich; South Dakota Republican Sen. Mike Rounds; and Indiana Republican Sen. Todd Young — aim to release a framework for moving ahead “in the near future.” (Spokespeople for the three other senators didn’t respond to requests for comment.)
But the future is quickly becoming the present. Congress may realistically have as little as six months to act before needing to break for August recess and, in many cases, return home to campaign for reelection, said Paul Gallant, a policy analyst at the market research firm Cowen Inc.
“That is not where I expected to be a year ago,” said Gallant. “A year later, we’re still having listening sessions and press conferences. Nothing looks likely to move.”
Initial momentum on AI regulation
For months, Congress has focused on getting up to speed on the basics of AI.
Last spring, as he kicked off what would become a global tour, OpenAI CEO Sam Altman left dozens of House lawmakers “riveted” at a private dinner in which he demonstrated ChatGPT’s capabilities. That same week, Altman made headlines at a Senate hearing by asking to be regulated — wowing senators who were more accustomed to tech CEOs dissembling or responding evasively to questions. (This week, Altman called for the creation of a global regulatory body for AI similar to the United Nation’s International Atomic Energy Agency.)
The momentum continued into the summer, as Schumer and the so-called AI gang including Heinrich, Rounds and Young organized three closed-door sessions to educate their colleagues on AI.
In June, Schumer announced still more learning sessions scheduled for the fall, as he outlined the fundamental principles that would drive the bill-making process. Any legislation on AI, he said, should prioritize US innovation first, but contain guardrails to preserve national security, promote transparency in AI products, protect election integrity and help Americans understand why an AI model behaves the way it does.
Schumer hosted nine closed-door panels for senators to hear from outside experts on AI, with the first one in September attracting a media frenzy due to the wide array of tech luminaries in attendance including Google CEO Sundar Pichai, OpenAI CEO Sam Altman and Nvidia CEO Jensen Huang.
Then things fell silent.
Proposals in the works
Congressional staff point to some signs of progress. Bipartisan discussions on AI have continued since the holidays on key Senate committees, the aide said. And House Speaker Mike Johnson wants to launch a bipartisan working group, potentially as soon as this month, to “make things happen” on AI, New York Republican Rep. Marcus Molinaro said at an event hosted by The Washington Post last week. Johnson reportedly met with Altman in January; a spokesperson for Johnson declined to comment.
Lawmakers have unveiled numerous proposals during this Congress, introducing more than 170 House and Senate bills in the past year alone mentioning artificial intelligence, according to a CNN review.
One bipartisan group of senators wants to ban deceptive AI deepfakes from US elections. Another calls for preventing AI from launching US nuclear weapons. A third bill would let artists, writers and everyday people sue companies for producing or hosting deepfakes of themselves, after songs and videos emerged featuring the artificial likenesses of musicians such as Drake and actors such as Tom Hanks. Still other ideas would require “high-risk” AI models to register for a government license, or create a dedicated new federal agency to oversee AI.
Expect the Senate Judiciary and Commerce committees to try to advance some of these measures this year, Tennessee Republican Sen. Marsha Blackburn said at a Washington event hosted by ITI, a tech industry trade group, last week.
And House leaders including Johnson and Minority Leader Hakeem Jeffries would like to pass 5 to 10 AI bills this year, said Virginia Democratic Rep. Don Beyer at State of the Net, another Washington conference, on Monday.
But despite the proliferation of ideas for how to rein in AI and the optimistic talk from lawmakers, Congress remains in a poised-for-action posture with no clear plan for getting those proposals to Biden’s desk.
A House Republican majority beset by infighting and a preoccupation with culture war issues makes it extraordinarily difficult to advance basic funding legislation, let alone complex artificial intelligence regulation, said a congressional aide on the House Energy and Commerce Committee.
“I don’t see how anything can get through the House in its current state,” the aide said. “You won’t get real, substantive change without the buy-in of both parties. And I think that’s impossible to do at this stage because too many members are not fully educated on AI and are not giving enough serious credence to the threat of AI as it relates to our elections.”
A limited window
If Congress cannot pass AI legislation this year, it can still try again next year, Gallant said. But it will mean leaving the 2024 elections exposed. And there are risks to extended inaction, as the tech industry barrels forward with AI development faster than policymakers can keep up.?And it could also cede the floor to other governments, including in the European Union, which is already poised to ratify a landmark regulation known as the EU AI Act.
“If they don’t pass something, it’ll be just like social media,” an industry that has long escaped regulation, Gallant said. “I suspect what it will take [to regulate AI] is a dramatic and potentially catastrophic, catalyzing event.”
Some of the deadlock could break if Congress finally reached a deal on a nationwide bill guaranteeing every American a fundamental right to digital privacy, said Alan Davidson, a senior Commerce Department official and a top advisor to the White House on telecommunications and technology issues. Last fall, Biden signed a sweeping executive order committing the executive branch to a rapid-fire timeline of tasks in an effort to get in front of AI issues, in what has further highlighted a contrast to Congress.
Many companies and consumer advocates describe a federal privacy bill as vital, low-hanging fruit that would help resolve some of the most basic policy questions around AI, including how data can be used or shared for training AI models.
“Comprehensive privacy legislation is something Congress could act on right now,” Davidson told CNN. “It would make a very big difference in terms of protecting people’s privacy and AI.”
Such a bill has been elusive for years, as Republicans and Democrats have split on the scope and reach of the specifics.
Absent congressional action, the task will fall to other government bodies — such as consumer protection agencies and financial regulators — to try to regulate AI under existing laws, said Sarah Myers West, managing director of the AI Now Institute and a former AI advisor to the Federal Trade Commission.
But, she added, many agencies need more funding to take on the added work, and executive agencies’ reach is increasingly under threat by court challenges, which could make the government’s job of protecting Americans from AI even harder — or dump the problem onto state legislatures.
