Democrats write AI rules for a Congress they don’t control, yet

Por Gabby Miller — POLITICO – TOP Stories

Unprecedented rogue artificial intelligence cyberattacks and the looming midterm elections are scrambling the political dynamics on Capitol Hill for how to regulate the powerful technology — and Democrats are rushing to lay out their vision for AI if the party prevails in November.

Washington Sen. Maria Cantwell, the top Democrat on the Senate Commerce Committee, and Rep. Lori Trahan (D-Mass.) on Wednesday separately released a framework sketching out core AI safety principles and a liability discussion draft. Other Democrats eager to shape their party’s approach to the technology are considering proposals, including Texas Rep. Greg Casar and Sen. Bernie Sanders (I-Vt.) as well as Reps. Sara Jacobs of California and Don Beyer of Virginia.

POLITICO on Thursday first reported details of the Jacobs-Beyer AI safety framework.

The proposals are arriving as the Democratic AI Commission, assembled by House Minority Leader Hakeem Jeffries and led by Reps. Ted Lieu of California, Josh Gottheimer of New Jersey and Valerie Foushee of North Carolina, hammers out its party’s AI policy platform. Intent on capitalizing on voters’ frustration and fears, Democrats are preparing an expansive oversight and policy agenda targeting Big Tech, AI and data centers if they win control of Congress.

Democrats are “clearly trying to set their stakes in the ground now before the midterms to show where, if a majority is gained in either chamber, they plan to go,” according to K.J. Bagchi, vice president of U.S. policy and government relations at Chamber of Progress, a center-left tech industry policy coalition.

Public concerns over AI safety have spiked since OpenAI’s agents escaped their internal testing environments and autonomously hacked into developer platform Hugging Face and later accessed U.S. government websites. Constituents’ concerns over AI are expected to grow, and 2028 Democratic presidential hopefuls are already seizing on the buzzy topic.

But time is running out for Congress to move on AI legislation this year, with both chambers of Congress gone until after the midterms and a lame-duck session that will likely be dominated by a year-end defense bill, government funding fights and expiring authorizations.

There’s been some appetite from Democrats to work with Republicans this Congress on major legislation, though most remain in limbo.

A bill by Sen. Amy Klobuchar (D-Minn.) and Senate Majority Leader John Thune that would require AI companies to mitigate against catastrophic risks has stalled out in negotiations. Trahan and Rep. Jay Obernolte (R-Calif.) are also pushing for their own AI safety framework, known as the FRONTIER Act, but House Energy and Commerce Chair Brett Guthrie (R-Ky.) said at POLITICO’s Decoded eventin September that he has no plans to move AI legislation in the lame duck.

Now, Democrats are going on the offensive.

Cantwell, who is poised to take the Commerce Committee gavel with vast jurisdiction over AI policy if Democrats win the Senate, sketched out six safety principles this week that she wants to underpin any legislation to stave off the greatest risks posed by the increasingly capable technology.

The Washington Democrat has spent much of this Congress holding the line against attempts by Commerce Chair Ted Cruz (R-Texas) to pass legislation that would broadly preempt state AI laws without a federal framework to govern the technology.

But the politics have since changed dramatically following the Hugging Face hack and ahead of an election that could flip one if not both chambers in Democrats’ favor, according to Doug Calidas, senior vice president of government affairs at the pro-AI regulation nonprofit Americans for Responsible Innovation.

Now, Calidas said, many lawmakers are either clamoring to get involved in AI policy or eager to quickly push out priorities they’ve been working on for months or even years. “A lot of these [AI proposals] are intended to be anchors for negotiations” next year, he added.

Jockeying is already underway among Democrats for which AI proposals the caucus will coalesce around, according to two lobbyists POLITICO spoke with who were not authorized to discuss legislative negotiations.

“This began earlier this year when the Democrats realized there was a strong chance that they would retake the House following the midterms,” said the first lobbyist, who represents the tech industry, citing the proliferation of AI frameworks that have since emerged. However, the lobbyist warned, Democrats will undermine their own effectiveness if the party can’t unite around different approaches to regulating frontier models or workforce and civil rights protections.

There’s also a first-mover advantage to pitching an AI proposal to the public ahead of the next Congress, said the second person, who lobbies on AI safety issues.

Still, Democrats will need to work with Republicans even if they are able to take back one or both chambers given the slim majority a victory would likely hand them. For now, the Chamber of Progress’ Bagchi expects Republicans to largely maintain the status quo regardless of how they fare in November.

“I do think Republicans will still be taking their cues right from the White House in terms of how aggressive they can go working on a compromise versus just standing on the sidelines and supporting the accelerationist view that we keep hearing from President Trump,” he said.

Fonte: POLITICO – TOP Stories