BURSTGreetings all,
Hope everyone’s hanging in. Going to try to keep this on the shorter side, given that it’s a holiday week and all. But I was seeing a number of celebrations (very justified) over what looks like a victory (and it is) but one that comes with an enormous caveat or three.
So the Senate passed Trump’s gargantuan “big beautiful bill,” and if it makes it through the House in roughly the same shape, it will, as many have noted, be a historic transfer of wealth from the poorest Americans to the richest. It’s the single biggest cut ($1 trillion) to Medicaid since the program’s inception, it cements ~$4 trillion worth of tax cuts for the wealthy, and it grants $45 billion in funding to ICE, making ICE the largest federal law enforcement agency. It cuts food assistance to the poor and scraps incentives for clean energy and electric vehicles. It adds $3 trillion to the US deficit. It’s a travesty.
If there’s a silver lining at all to be found here, it’s that the 10-year ban on state-level lawmaking on AI was stripped from the budget bill mere hours before it passed. “Early Tuesday morning, the United States Senate voted 99-1 to pass an amendment to the budget bill removing the proposed 10-year moratorium on the enforcement of state laws on artificial intelligence,” as the Tech Policy Press, which has been doggedly following the bill, noted.
Big tech stocks tumbled when the news broke.
A quick note that, as always, 100% of this work is made possible by paid subscribers—if you find value in this analysis and reporting, and you’re able, please consider subscribing or upgrading. Infinite thanks to all who already do. Now, onwards.
I reported on this project at length in May, and spoke to California state rep Isaac Bryan, who’d written one of the state AI laws that the GOP’s amendment aimed to kill. It is unambiguously a good thing that it’s dead, and that Silicon Valley’s power is not ‘halt states’ capacity to do democracy with regard to our products’ levels of absolute quite yet. But it’s close. The state AI law ban really was as draconian and anti-democratic as it sounded. In its original form, it read “no State or political subdivision thereof may enforce any law or regulation regulating artificial intelligence models, artificial intelligence systems, or automated decision systems during the 10-year period beginning on the date of the enactment of this Act.”
But now that it’s dead, and we’re all getting swept up in the swirl of well-justified outrage over the amendments that *did* pass and the spotlight turns to the House, I worry many will forget what almost happened here. That Silicon Valley elites almost won a battle to stop states from passing AI laws, period.
Ultimately, they were one vote short. But just one.
I would suggest that we do not forget that OpenAI, Meta, Google, and some of the top VC firms in Silicon Valley were behind this bill and driving for its passage. Or that, had these companies not sent teams of lobbyists to DC and sold the idea that AI is a zero sum arms race with China, it would not have existed at all. That biggest AI companies—Anthropic excepted, to its credit, even if its opposition was performative—were directly campaigning to exempt their products and practices from the democratic process. They were actively pushing to radically de-democratize a technology they are selling us all as a great force for democratization.
As OpenAI and Meta et al quietly move on to building and selling their next generation of AI products, don’t forget what they tried to do. Which, essentially, can be boiled down to “stop California from passing laws to regulate AI that might take a little time and money to comply with.”
Here’s Politico to remind us of that campaign:
As California and other states move to regulate AI, companies like OpenAI, Meta, Google and IBM are all urging Washington to pass national AI rules that would rein in state laws they don’t like. So is Andreessen Horowitz, a Silicon Valley-based venture capitalist firm closely tied to President Donald Trump.
Tech lobbyists are pushing the White House to oppose state AI laws, sitting down with key lawmakers to emphasize the threat and — perhaps most crucially — huddling with fellow lobbyists to see if they can unite behind a national strategy.
Their effort puts the companies in the unusual position of supporting federal regulation on the fast-moving technology, which they have so far managed to stave off almost entirely.
The tech industry has recognized that because federal regulation of AI is all but off the table now, the only real risk that their products will be regulated comes from states like California, New York, and Colorado, where more progressive legislatures are able to pass meaningful laws that can impact the market for tech products. This was an effort to shut that down wholesale.
And it looked like it was going to pass, right up until the last minute, when Marsha Blackburn withdrew her support for the second time in short succession. The ban had already been sanded down to 5 years, and would have functioned by tying it to states’ eligibility to receive broadband funding. If a state passed any AI laws, the federal government would cut off broadband funds. But Blackburn is a senator of Tennessee, where country music artists in Nashville had pushed for protections from AI, and the state responded by passing ELVIS—the Ensuring Likeness Voice and Image Security Act. Blackburn worried the AI ban would interfere with its implementation, and with lawmakers’ ability to write laws to protect children from digital products and services.
Opposition from civil society groups, Democrats and other Republicans had already muddied the waters, and there was opposition in other corners—but when Ted Cruz, one of the most vocal backers of the AI law ban, cut a deal with Blackburn exempting her state’s law, it looked like the moratorium would pass. Eleventh hour calls from Steve Bannon, of all people, caused her to switch again, and when she announced she was out, everyone else abandoned the measure too.
That’s why the measure officially failed by 99-1, as once-supportive senators switched their votes to save face as it became clear the amendment would fail, but it really was just one vote shy of passing. Let’s not forget that. Because yes, sure, any industry coalition that sees a shot at exempting itself from costly regulation is probably going to try to take it—but most industries aren’t promising wide-scale social transformation or that they are going to rewrite the social contract. We must take note of what has happened here, and henceforth regard all of Sam Altman’s talk of the need for careful regulation or for democracies to adapt to AI as the pablum it has always been. We must understand that these companies have made a direct play for—and will continue to seek—absolute control over their automation and surveillance technologies, no matter the social cost.
Cruz has floated the idea of passing a state AI law ban as a standalone bill, so there’s little doubt that we haven’t seen the last of such efforts. Silicon Valley continues to hold sway over the Trump administration, and vice versa. But for now, state level lawmaking can persist. One of California’s bills, “No Robo Bosses,” which seeks to prevent algorithmic discrimination in the hiring process, among other things, just cleared a hurdle in the state legislature—it’s one step closer to become law.
The states are the laboratory of democracy, as the GOP likes to remind us. The prime movers in AI were willing to shut it down.
More bloody bits
Peter Lewis, writing in the Guardian and citing my book (cheers Peter), calls on Australians to embrace their inner Luddite—many convicted machine breakers were “transported” to Australia as punishment for their luddism in the 1810s—when considering AI.
From the piece, which is well worth a read:
If I hear another well-intentioned person justifying their support for the regulation of AI with the qualifier “I’m no luddite, but …” I’m going to start breaking my own machine.
From ministers to union leaders to progressives watching from the cheap seats, there is growing recognition that untrammelled development of this technology carries significant risks.
But there is also a reticence to be seen as being anti-technology lest we are perceived as standing in the way of the productivity boom and consequent bounty of abundance that the boosters of these tools promise is just around the corner. After all, we aren’t luddites.
The problem with being forced into this defensive mindset is that we misread the challenge at hand, which is not so much about the nature of the technology but the power dynamics driving this change.
Read the whole thing in the Guardian.
-In a thoughtful and far-ranging conversation about what socially mediated AI products are doing to our ability to discern reality with New Yorker writer Jia Tolentino, on his Why Is This Happening? podcast, Chris Hayes gave a surprise shout to this very newsletter. Thanks for the nod, and hello to any readers who found me there. Welcome aboard—and hammers up.
That’s it for today. More soon; stay frosty out there.