# ‘Claude, How Should I Vote in 2026?’

An episode of [Big Take](https://invisiblenetwork.ai/podcast/1578096201), 1 Jul 2026, 17 min.

Audio: https://podtrac.com/pts/redirect.mp3/tracking.swap.fm/track/UVBrz8bN8aM2Xe47PEPu/traffic.omny.fm/d/clips/e73c998e-6e60-432f-8610-ae210140c5b1/825d4e29-b616-46f4-afd7-ae2b0013005c/c93922b2-4086-45c9-ab1a-b47a013629f3/audio.mp3?utm_source=Podcast&in_playlist=8b1dd624-a026-43e9-8b57-ae2b00130066

In the 2026 elections, artificial intelligence is the backdrop for just about everything. The US and Anthropic reached a truce in recent days that lets the company roll out its powerful models to some users, the sort of on-the-fly decision-making that has candidates jockeying to shape how AI is regulated. Big AI companies are pouring money into 2026 state and federal races, dwarfing the cryptocurrency industry’s spending in 2024. And even as AI is fueling Americans’ anxiety about jobs and energy prices, it’s feeding them the ads and information that will shape their votes: According to one survey, 55% of US voters under the age of 45 are likely to use chatbots to learn about candidates and elections. On today’s Big Take podcast, Bloomberg’s Michael Shepard, a senior editor focusing on AI, joins David Gura to discuss what the first AI election means for our current moment. We have a special Bloomberg subscription offer for podcast listeners at Bloomberg.com/podcastoffer. Hosted by David Gura; Produced by Rachael Lewis-Krisky; Reported by Michael Shepard; Edited by Jeffrey Grocott. Fact-checking by Victor Swezey, Julia Press and David Fox; Engineering by Emma Munger. Senior Producer: Naomi Shavin; Deputy Executive Producer: Julia Weaver. Executive Producer: Nicole Beemsterboer. See omnystudio.com/listener for privacy information.

## Transcript

[0:02] Speaker 1: Bloomberg Audio Studios, podcasts, radio news. It's really coming at
[0:10] Speaker 1: us all at once this year, in this election cycle.
[0:14] Speaker 2: Artificial intelligence is on the ballot in many twenty twenty
[0:17] Speaker 2: six midterm elections. How to regulate it, what it means
[0:20] Speaker 2: for jobs and for energy prices? And hundreds of millions
[0:23] Speaker 2: of dollars are pouring into election races from competing super
[0:26] Speaker 2: PACs vuying to steer the AI conversation. Voters are getting
[0:30] Speaker 2: more and more election information from AI, and they're seeing
[0:34] Speaker 2: campaign ads made with it. As one former US official
[0:37] Speaker 2: recently told Bloomberg, this is the first AI election.
[0:41] Speaker 1: We're seeing it in the campaign financing, we're seeing it
[0:44] Speaker 1: in AI is an issue itself in the way that
[0:48] Speaker 1: people are now becoming truly acutely aware of its impact
[0:52] Speaker 1: on the economy, and it's also becoming a campaign tool.
[0:57] Speaker 2: Michael Shepherd is a Bloomberg Senior editor for Technology and
[1:00] Speaker 2: Strategic Industries who focuses on artificial intelligence.
[1:03] Speaker 1: It's rewriting the playbook for how candidates seek office and
[1:07] Speaker 1: the strategies that they can use to gain an edge
[1:10] Speaker 1: in competitive races.
[1:11] Speaker 2: Mike says, the concerns about AI that are coming out
[1:14] Speaker 2: this election season mirror the ones across our society at
[1:18] Speaker 2: a moment when AI can tell us just about anything
[1:21] Speaker 2: short of where the AI revolution will take us. This
[1:28] Speaker 2: is the big take from Bloomberg News. I'm David Gera
[1:30] Speaker 2: on today's episode the first AI Elections. What's at stake
[1:34] Speaker 2: and who's working on the answers? Mike, Let's start with
[1:42] Speaker 2: campaign advertising. How has AI changed the kind of ads
[1:46] Speaker 2: that the people see?
[1:47] Speaker 1: This is, actually, David, a great place to start because
[1:50] Speaker 1: this is one of the most visible areas for voters
[1:53] Speaker 1: in the public to see AI shaping the elections. Candidates
[1:58] Speaker 1: are deploying it actively on the campaign trail. And in
[2:02] Speaker 1: the past, to get an advertisement or a message out,
[2:05] Speaker 1: you would have to get a radio spot, a newspaper ad,
[2:08] Speaker 1: very expensive television advertising, and you know, for candidates that
[2:12] Speaker 1: were on a shoe string budget or just starting out
[2:15] Speaker 1: or didn't have a name, that was tough. But AI
[2:17] Speaker 1: is enabling them to create images and curate them very quickly.
[2:22] Speaker 1: It's also allowing candidates to develop deep fake ads, you know,
[2:26] Speaker 1: some that portray candidates in unfavorable lights to gain it
[2:31] Speaker 1: edge in a race.
[2:32] Speaker 2: I got to imagine it's changing the way that voters
[2:35] Speaker 2: research candidates learn about candidates as well, which.
[2:38] Speaker 1: Sure is David, because what we see is the same
[2:41] Speaker 1: thing with politics that we see elsewhere in the economy.
[2:45] Speaker 1: And there's some research by a progressive aligned group called
[2:49] Speaker 1: Data for Progress that indicates that, you know, voters under
[2:52] Speaker 1: the age of forty five are more likely to use
[2:55] Speaker 1: AI as a tool to find out about candidates. And
[2:58] Speaker 1: what that suggests also so, is that AI is a
[3:01] Speaker 1: tool for political research.
[3:04] Speaker 2: Is here to say, Mike and your role helping to
[3:07] Speaker 2: lead Bloomberg's coverage of AI, you have a special vantage.
[3:10] Speaker 2: You're based in Washington, so you're looking at the way
[3:12] Speaker 2: these companies are trying to influence politics and the political conversation.
[3:18] Speaker 2: This is an election in which a lot of these
[3:19] Speaker 2: companies and their affiliates are pouring a lot of money
[3:22] Speaker 2: into these races. What does that look like in broad terms?
[3:25] Speaker 1: On broad terms, it looks like a lot of cash.
[3:28] Speaker 1: And you know, to give you an idea, the pledges
[3:32] Speaker 1: so far that we've gathered from across the industry amount
[3:36] Speaker 1: to two hundred and seventy five million across the US,
[3:41] Speaker 1: and you know that, if fulfilled, would dwarf the amount
[3:46] Speaker 1: that the crypto industry spent during the twenty twenty four election,
[3:51] Speaker 1: when the digital currency industry was able to topple a
[3:55] Speaker 1: number of Democratic incumbents, and the AI industry is certainly
[3:59] Speaker 1: trying to repeat that same playbook to its advantage.
[4:04] Speaker 2: Mike, I want to ask you about one race, in particular,
[4:06] Speaker 2: a primary race that took place here in New York
[4:09] Speaker 2: City involving Alex Boris. He was one of the candidates
[4:12] Speaker 2: running in this primary, running to be the Democratic candidate
[4:14] Speaker 2: in the twelfth Congressional district, which includes a big part
[4:17] Speaker 2: of Manhattan. Alex Boris lost that primary, but there was
[4:21] Speaker 2: so much money in this one race, and I should
[4:24] Speaker 2: say Michael Bloomberg, the owner of Bloomberg LP, the parent
[4:26] Speaker 2: company of Bloomberg News, endorsed and contributed to the winning candidate,
[4:30] Speaker 2: Michael Lasher's campaign. But there was a lot of AI
[4:33] Speaker 2: money in this race as well. What about it attracted
[4:36] Speaker 2: so many big backers.
[4:38] Speaker 1: Well, it was Alex Boris really who drew all of
[4:41] Speaker 1: that attention, and that was because as a New York
[4:44] Speaker 1: State Assemblyman, he had sponsored a number of bills that
[4:48] Speaker 1: sought to reign in the industry. Governor hokel just signed
[4:52] Speaker 1: my raiz Act.
[4:53] Speaker 3: New York has the strongest AI safety law in the country.
[4:57] Speaker 1: He pushed safety legislation. He also pushed a data center
[5:01] Speaker 1: moratorium that ultimately passed, And so he came into this
[5:06] Speaker 1: race with kind of a target on his back for
[5:09] Speaker 1: a number in the industry who see the kinds of
[5:12] Speaker 1: bills that he was proposing and the kinds of policies
[5:15] Speaker 1: that he was advocating as a threat to their attempt
[5:19] Speaker 1: to build out the industry at a breakneck pace. I mean,
[5:22] Speaker 1: if people think I've been too friendly to tech, someone
[5:25] Speaker 1: should tell the executives at palent here that are funding
[5:27] Speaker 1: the superpack against me. And so, as a result, Alex
[5:31] Speaker 1: Boris ended up drawing a lot of attention but also
[5:33] Speaker 1: a lot of money into this race. We estimate that
[5:37] Speaker 1: twenty nine million dollars were spent by all sides in
[5:40] Speaker 1: industry in the competition. Of course, he ultimately lost to
[5:45] Speaker 1: the candidate was favored to win, but it still drew
[5:48] Speaker 1: a significant attention to the issue and made New York
[5:51] Speaker 1: twelve kind of a proxy battle for the industry.
[5:55] Speaker 2: What's so fascinating to me is there was AI money
[5:58] Speaker 2: on both sides of this race. Your team just reported
[6:00] Speaker 2: that a super pac called Leading the Future spent about
[6:03] Speaker 2: eight million dollars opposing Boris. That super pac supports softer
[6:08] Speaker 2: regulation on AI, and it's funded by an Open Ai
[6:10] Speaker 2: co founder, among others. That on the other side, fifteen
[6:13] Speaker 2: million dollars came into support Boris from Public First Action,
[6:16] Speaker 2: whose funders include Anthropic Why were anthropics interests kind of
[6:21] Speaker 2: aligned with Boris's interests and what have you seen from
[6:24] Speaker 2: other companies in that regard.
[6:26] Speaker 1: Well, I'm glad you brought that up, David, because it
[6:28] Speaker 1: really does highlight that silicon value is not of one
[6:31] Speaker 1: mind when it comes to AI. There are a lot
[6:33] Speaker 1: of people in the industry who also favor some restraint
[6:37] Speaker 1: on the technology, and ANTHROPICCA is one of the leading voices.
[6:41] Speaker 1: In fact, part of its branding is that it professes
[6:44] Speaker 1: itself as a more responsible AI developer, one that places
[6:49] Speaker 1: a higher priority perhaps and its competitors when it comes
[6:52] Speaker 1: to safety. So they naturally leaned toward a candidate who
[6:56] Speaker 1: would favor more regulation, and that was.
[6:59] Speaker 3: Alex for is that race.
[7:01] Speaker 2: The New York twelfth primary might have been an outlier
[7:03] Speaker 2: in terms of how much money was poured into it,
[7:06] Speaker 2: but stepping back, I wonder sort of how much of
[7:08] Speaker 2: an outlier it is in terms of just the way
[7:10] Speaker 2: that AI is a campaign issue going forward here, I
[7:13] Speaker 2: think about the kind of broader zeitgeist, how people feel
[7:17] Speaker 2: about AI, its prospects, its potential perils. How does that
[7:21] Speaker 2: manifest in other races across the country.
[7:23] Speaker 1: Well, it's a manifesting itself, not so much as like
[7:26] Speaker 1: the headline issue, Like people are not voting up and
[7:29] Speaker 1: down on AI per se, but they are wrapping AI
[7:33] Speaker 1: into all sorts of other questions, principally affordability, because when
[7:38] Speaker 1: you think about affordability, all you have to do is
[7:40] Speaker 1: look at your electricity bill if you live in certain
[7:42] Speaker 1: parts of the country and see that, yes, it has
[7:45] Speaker 1: gone up. We had a big take recently that looked
[7:50] Speaker 1: at how electricity prices had more than tripled in some areas,
[7:54] Speaker 1: up two hundred and sixty seven percent, as a result
[7:56] Speaker 1: of all the AI data centers that are being built
[7:59] Speaker 1: with all the investment from Silicon Valley. So people are
[8:02] Speaker 1: looking at that and thinking, hey, you know, this is
[8:04] Speaker 1: hitting my pocketbook. I feel it for real. They're also
[8:07] Speaker 1: worried about their jobs. Will artificial intelligence come in somehow
[8:12] Speaker 1: and change the way I do my job on a
[8:15] Speaker 1: daily basis, or perhaps eliminate it all together, or maybe
[8:19] Speaker 1: eliminate that entry level opportunity that might be awaiting my
[8:22] Speaker 1: son or daughter because that kind of lower level maybe
[8:26] Speaker 1: grump work that was good for us when we were
[8:29] Speaker 1: starting out, could now be done by a machine. And
[8:31] Speaker 1: then there is also just the distrust of Silicon Valley
[8:35] Speaker 1: because policymakers here in the US have never really gotten
[8:38] Speaker 1: around to resolving some of the key questions surrounding social
[8:42] Speaker 1: media and how it should be overseen. And AI is
[8:45] Speaker 1: an extension of that in a way in many voters' eyes,
[8:49] Speaker 1: and they are concerned that the government is not going
[8:52] Speaker 1: to be ready to step up here, and polls are
[8:55] Speaker 1: showing and we identify this in our big take that
[8:59] Speaker 1: they I think that the government needs to do.
[9:01] Speaker 2: More regulation isn't keeping up with advancements in AI, and
[9:09] Speaker 2: in the breach, the US government and companies like Anthropic
[9:12] Speaker 2: are laying down rules as they go. That's next, Mike,
[9:24] Speaker 2: You just went through a lot of the concerns that
[9:26] Speaker 2: people have about AI, how it's going to affect their
[9:28] Speaker 2: jobs and their family finances, their communities, their privacy, their security.
[9:32] Speaker 2: I know you've been digging into how these questions have
[9:34] Speaker 2: been playing out with one company in particular that's Anthropic,
[9:38] Speaker 2: and anthropic stances brought it into conflict with the US government.
[9:41] Speaker 2: Things really blew up earlier this year one Anthropic effectively
[9:44] Speaker 2: set guardrails on how the Defense Department the Pentagon could
[9:48] Speaker 2: use its technology, and the Trump administration hit back saying
[9:51] Speaker 2: it would designate Anthropic a national security risk and effectively
[9:55] Speaker 2: blacklist that company from being a government supplier.
[9:59] Speaker 3: What's the latest on this fight.
[10:01] Speaker 1: Well, David, there was another conflict that erupted in early
[10:04] Speaker 1: June between the company and the Trump administration and it
[10:08] Speaker 1: centered on Anthropics top two models, Fable five and Mythos five.
[10:12] Speaker 1: In essence, the government was warning the company that it
[10:15] Speaker 1: needed to contain what officials saw as a so called jailbreak,
[10:19] Speaker 1: and that refers to an ability to get around the
[10:21] Speaker 1: safety and security guardrails. And if those concerns weren't addressed,
[10:26] Speaker 1: the company would need export licenses from the Commerce Department
[10:30] Speaker 1: to be able to share that technology with any foreign national,
[10:33] Speaker 1: and that included any foreign person on US soil. In effect,
[10:38] Speaker 1: that forced Anthropics hand the company had to immediately disable
[10:42] Speaker 1: access to those two models starting on June twelfth until
[10:46] Speaker 1: the security concerns of the government were addressed. Now at
[10:49] Speaker 1: the end of June, if we fast forward a bit
[10:52] Speaker 1: Commerce Secretary Howard Luten, it goes back to the company
[10:55] Speaker 1: to say, look, you have now taken steps to address
[10:58] Speaker 1: our concerns and in a fe we will no longer
[11:01] Speaker 1: require export controls on those models. The move and effect
[11:05] Speaker 1: has allowed Anthropic to begin releasing Fable five as originally
[11:11] Speaker 1: intended to a broad group of users globally without any limits.
[11:16] Speaker 1: Mythos five, however, has been restricted still to just US
[11:20] Speaker 1: users and on a more limited basis. And you'll remember,
[11:23] Speaker 1: Mythos is that cyber capable model that everybody had really
[11:28] Speaker 1: wanted to be able to use to test their own
[11:30] Speaker 1: networks for possible vulnerabilities that could be exploited by hackers.
[11:35] Speaker 1: And you know, David, this whole episode does raise some
[11:38] Speaker 1: broader questions too, and in essence they boil down to
[11:41] Speaker 1: where the government is heading when it comes to AI regulation.
[11:45] Speaker 1: The Trump administration really from the start has insisted would
[11:49] Speaker 1: take a hands off approach to AI, especially in the
[11:53] Speaker 1: area of regulation. However, it is unfolding on a bit
[11:56] Speaker 1: of an ad hoc basis as we are seeing with
[11:59] Speaker 1: its handling of Anthropic and those two models. And what's
[12:03] Speaker 1: unclear is what the roadmap is ahead for Anthropic and
[12:06] Speaker 1: its rivals Open Ai and Google's Deep Mind as they
[12:10] Speaker 1: start to roll out equally sophisticated models, will they have
[12:15] Speaker 1: to go through the same sorts of checks before the
[12:18] Speaker 1: government gives their clearance for release. It is the very
[12:22] Speaker 1: kind of thing that Silicon Valley companies have been worried
[12:25] Speaker 1: about and warning against.
[12:27] Speaker 2: Your team at Bloomberg just had an exclusive report that
[12:29] Speaker 2: Anthropic raised an alarm of its own about Chinese users.
[12:33] Speaker 3: Walk us through what happened.
[12:35] Speaker 1: We were privy to a letter and we're told about
[12:39] Speaker 1: a letter that the company had sent to several US
[12:43] Speaker 1: senators and White House officials warning that Ali Baba, which
[12:47] Speaker 1: is one of the biggest tech companies in China, have
[12:51] Speaker 1: been improperly extracting data from the company's Claude model. And
[12:57] Speaker 1: Claude is really the centerpiece of Anthropics product line. It
[13:02] Speaker 1: is something that we have seen adopted across the economy
[13:06] Speaker 1: by businesses and even consumers are using it widely. But
[13:09] Speaker 1: what Anthropic is saying is that Ali Baba and operators
[13:13] Speaker 1: affiliated with it used tens of thousands of accounts to
[13:20] Speaker 1: make nearly thirty million queries to claud in a bid
[13:24] Speaker 1: to take those results and then build its own model,
[13:29] Speaker 1: one that would compete with Anthropics, but one that would
[13:33] Speaker 1: also not carry the same degree of safety guardrails. This
[13:37] Speaker 1: is a process called distillation. And I don't know how
[13:40] Speaker 1: wonky we want to get here, David, but you're going
[13:43] Speaker 1: to start hearing more about this word because Anthropic and
[13:47] Speaker 1: other top USAI labs have been warning about distillation all
[13:52] Speaker 1: year long. Now. Distillation itself is not illegal, and a
[13:56] Speaker 1: lot of companies even allow it and encourage it to
[13:59] Speaker 1: a certain level. But the issue is when they are
[14:02] Speaker 1: extracting data at an industrial and systemic scale and then
[14:06] Speaker 1: using it to develop a rival frontier model. That is
[14:10] Speaker 1: when a line is crossed. In Anthropics view and also
[14:14] Speaker 1: in the view of companies like open AI.
[14:16] Speaker 3: I want to pull back.
[14:17] Speaker 2: And clearly there's a lot at stake here in terms
[14:20] Speaker 2: of what kind of role AI plays in our lives,
[14:22] Speaker 2: international security, in our economy. Just a ton of money
[14:26] Speaker 2: riding on AI in global markets with big IPOs coming
[14:29] Speaker 2: down the Pike pretty soon here. What is that funding
[14:32] Speaker 2: going to allow them to do well?
[14:34] Speaker 1: The funding is really everything. They need That money to
[14:38] Speaker 1: keep building AI is I don't want to call it
[14:41] Speaker 1: the perpetual machine, but they really are investing in the
[14:46] Speaker 1: infrastructure that they need to keep running those models. And
[14:49] Speaker 1: it's expensive. Those chips from Nvidia can be tens of
[14:55] Speaker 1: thousands of dollars a piece from AMD their expense, and
[15:00] Speaker 1: they need those on a scale that it's really hard
[15:03] Speaker 1: to describe in a podcast or radio forumat you really
[15:06] Speaker 1: have to see the pictures of these data centers just
[15:09] Speaker 1: to get an idea of the footprint that they occupy
[15:13] Speaker 1: and also the amount of computing power that they deliver.
[15:17] Speaker 1: And if you think about all the businesses across the
[15:20] Speaker 1: economy and as they try to adopt AI and employ it, well,
[15:23] Speaker 1: think of how much more computing power would be needed
[15:27] Speaker 1: to make that happen.
[15:28] Speaker 2: Mike, let me close by asking you just when we
[15:30] Speaker 2: think about where we go from here, who are what
[15:32] Speaker 2: has the most power to kind of influence the conversation
[15:34] Speaker 2: about how we're going to steer our adoption of AI.
[15:38] Speaker 2: What are the odds that in two years time, if
[15:41] Speaker 2: you and I were to sit down. The general election
[15:43] Speaker 2: of twenty twenty eight would also be the AI election.
[15:48] Speaker 1: David, I am going to mark the calendar for two
[15:51] Speaker 1: years from now, because yes, we will have this conversation again.
[15:56] Speaker 1: The stakes are only going to grow bigger because you know,
[16:00] Speaker 1: whoever is to succeed Donald Trump as the candidate for
[16:03] Speaker 1: the Republican nomination, whoever is vying for the Democratic nomination,
[16:09] Speaker 1: they will have to address these issues. And if there is,
[16:12] Speaker 1: for example, a market correction or a pullback, or some
[16:15] Speaker 1: other negative externality related to artificial intelligence, that too will
[16:20] Speaker 1: be on the ballot. And if lawmakers and the administration
[16:24] Speaker 1: really can't get it together when it comes to regulation,
[16:27] Speaker 1: the conversation will only intensify. So you can count on it.
[16:32] Speaker 1: Twenty twenty eight, AI will once again be on the ballot.
[16:35] Speaker 1: It'll be part of the conversation politically and really for
[16:38] Speaker 1: the foreseeable future.
[16:44] Speaker 2: This is the Big Take from Bloomberg News. I'm David Gerat.
[16:46] Speaker 2: To get more from The Big Take and unlimited access
[16:49] Speaker 2: to all of Bloomberg dot com, subscribe today at Bloomberg
[16:52] Speaker 2: dot com.
[16:52] Speaker 3: Slash podcast offer.
[16:54] Speaker 2: If you like this episode, make sure to follow and
[16:56] Speaker 2: review The Big take wherever you listen to podcasts, it
[16:59] Speaker 2: helps people find the shit out.
[17:00] Speaker 3: Thanks for listening. We'll be back tomorrow. M

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.