Onpode
Cover art for Anthropic just cut Claude access in half for California government workers — here's the deal

Anthropic just cut Claude access in half for California government workers — here's the deal

June 29, 2026 · 5 min

Cole Brennan & Malcolm Reeves

California gave all state agencies access to Anthropic's Claude AI at 50% off, announced June 29 with Governor Newsom's name on it — despite an active federal supply-chain risk designation on Anthropic. The California DMV was already quietly running Claude for customer service before the announcement legitimized it statewide.

On June 29, 2026, California Governor Gavin Newsom announced a first-of-its-kind statewide partnership with AI company Anthropic, making Claude the first AI productivity tool available to all California state agencies and local governments — including cities and counties — at a 50% discount off standard enterprise pricing.

0:005:12
Make your own on Onpode

Describe any topic. Hear it in minutes.

More Onpode episodes on Technology

About this episode

Before Governor Newsom announced his statewide deal with Anthropic, the California DMV was already quietly using Claude for customer service. That detail sits at the center of this episode — and it reframes everything else. Was June 29th a bold federalism moment, or just the legitimization of something that had already happened? The deal gives every California state agency access to Claude through the Department of Technology's shared services portal at 50% off. The Trump administration had designated Anthropic a supply-chain risk and the Pentagon had restricted its most advanced models at the federal level. California moved anyway, publicly, with Newsom's name on it. The episode digs into what that authority actually means — and where it breaks down. Newsom's own language promises that AI won't replace government workers, but the accountability structure is fragmented: each agency evaluates deployment individually, no one has a mandate to track workforce impact statewide. The free training Anthropic provides sounds like a safeguard; the episode asks what exactly workers are being trained to do once entry-level tasks are absorbed. The federal supply-chain flag hasn't moved. So either the federal government responds and California becomes a cautionary tale, or it doesn't and every state now has a template. Both outcomes set a precedent — and the episode is honest that it can't tell you which one comes next.

Frequently asked

What is the California Anthropic Claude deal?

California's June 29 deal makes Anthropic's Claude the first AI productivity tool available to all state agencies through the Department of Technology's shared services portal at a 50% discount. Governor Newsom announced it publicly, with free workforce training from Anthropic developers included. The California DMV was already using Claude before the announcement.

Did California override a federal restriction on Anthropic's Claude?

Yes. The Trump administration placed a supply-chain risk designation on Anthropic and the Pentagon restricted its most advanced models at the federal level. California's Claude deal, signed June 29 with Governor Newsom's name on it, directly contradicts that federal determination — a move the transcript describes as California daring the federal government to respond.

Was California already using Claude AI before the Newsom announcement?

Yes. The California DMV was running Claude for customer service before June 29, before the statewide announcement, and before Claude was added to the shared services portal by Chris Given. The June 29 announcement did not introduce Claude to California government — it legitimized deployment that was already underway.

Who is accountable if California government agencies misuse Claude AI?

Under the California-Anthropic deal's structure, each agency is individually responsible for accuracy, transparency, and privacy — there is no centralized accountability body. The LA Times flagged that adequate guardrails for data privacy are not in place at this scale, meaning oversight exists only at the agency level, with no statewide mandate to track aggregate impacts.

Could the California Claude deal affect government jobs?

The deal includes free workforce training from Anthropic developers, but critics argue the training prepares workers to supervise AI rather than replacing the entry-level roles Claude absorbs. No statewide mechanism exists to track whether agency headcount reductions connect to Claude deployments — that causal link has no one with a mandate to monitor it.

Grounded in 10 sources
Anthropic partners with California to expand AI use by government workers - Los Angeles Times · latimes.com
Exclusive: Newsom, Anthropic ink deal to expand government use - Politico · politico.com
Newsom, Anthropic Ink Deal to Expand Government Use - Business Insider · businessinsider.com
Anthropic and Gov. Newsom forge deal allowing California government to use Claude at half price - TechCrunch · techcrunch.com
Anthropic and Gov. Newsom forge deal allowing California ... · techcrunch.com
California gives all state agencies access to Claude at half price in Anthropic deal · thenextweb.com
California Snubbed the Pentagon's AI Blacklist - and Got Claude at Half Price - Gadget Review · gadgetreview.com
Governor Newsom announces a first-of-its-kind partnership ... · gov.ca.gov
OpenAI and Anthropic Limit New AI Models to Trump-Approved Customers During Cybersecurity Review - SecurityWeek · securityweek.com
California agencies get access to Anthropic's AI tools at half price | StateScoop · statescoop.com
Read transcript

Cole Brennan: Hey — okay, something's been bugging me since last week and I genuinely need to talk it through, because I cannot decide if this California-Anthropic deal is a big constitutional moment or just a governor helping out a hometown company.

Malcolm Reeves: I've been waiting for this one, yes — though I'd push back on the framing slightly. Those two things aren't mutually exclusive.

Cole Brennan: Right, that's the thing — they're not. But like... the Trump administration put a supply-chain risk designation on Anthropic. The Pentagon restricted its most advanced models at the federal level. And Newsom signs the deal anyway, on June 29th, publicly, and calls it the California way. That's not quiet.

Malcolm Reeves: It's not quiet at all. And the plainest way I can put it — California looked at a federal security determination and decided the state had the authority to reach a different conclusion. That's the core of it, before anything else.

Cole Brennan: So who actually has that authority? Like, is there a clear answer?

Malcolm Reeves: That's the question California just forced everyone to answer. And they forced it by making Claude the first AI productivity tool available to all California state agencies through the Department of Technology's portal — fifty percent off, statewide, before anyone could stop them.

Cole Brennan: But wait — San Francisco already did this. Microsoft 365 Copilot Chat, nearly thirty thousand city employees, before June 29th, before any of this. So California isn't even the first mover here. They're second. And the one getting all the attention — the one Anthropic leaked exclusively to Politico before the public even knew — happens to be a San Francisco company calling it a 'home-state responsibility.' I mean, some people would say that's not federalism. That's just... backing your own.

Malcolm Reeves: The DMV.

Cole Brennan: What?

Malcolm Reeves: The California DMV was already running Claude for customer service before June 29th. Before the announcement, before Chris Given put it in the shared services portal, before Newsom called it the California way — the infrastructure was there. Now, that's not procurement. That's not a governor helping out a hometown company. That's a deliberate end-run while the federal government's back was turned. The announcement didn't introduce Claude to California government. It legitimized what was already happening.

Cole Brennan: Okay but — no, actually, that makes my point worse, not better. If the DMV was already quietly using a tool the Pentagon flagged as a supply-chain risk, and nobody stopped it, and then Anthropic gets an exclusive with Politico and suddenly it's a statewide deal at fifty percent off... I mean, that's not a principled stand against federal overreach. That's just getting caught and calling it a policy.

Malcolm Reeves: That's one read. But California overriding the Pentagon's designation publicly, with Newsom's name on it, is not the move of someone trying to hide anything. That's the move of someone daring the federal government to respond.

Malcolm Reeves: But daring them to respond isn't the same as having a plan if they do. And that's where Newsom's language starts to crack. 'AI should not replace the human work of government' — that's a quote, that's his actual statement. Now, agencies are individually responsible for accuracy, transparency, privacy. Each one. No centralized accountability. So who's watching the whole picture?

Cole Brennan: Nobody, technically.

Malcolm Reeves: Nobody. Case-by-case deployment evaluation per agency. That's the actual structure. The LA Times flagged it — adequate guardrails for data privacy aren't in place at this scale. So the promise lives at the press release level, and the accountability lives... nowhere centralized.

Cole Brennan: Okay but — wait, I want to defend the Tuesday morning version of this. A contracts analyst at General Services, Claude drafts a vendor agreement summary in ninety seconds, she fact-checks it, saves forty minutes a day. That's real. That's better government, right?

Malcolm Reeves: Follow it to year three.

Cole Brennan: I mean — yeah, I know where you're going. The junior analyst hire that never happens next budget cycle. But the deal includes free workforce training from Anthropic developers, that's built in. That's not nothing.

Malcolm Reeves: Training to do what, though? Training to supervise the thing that absorbed the entry-level work? That's not augmentation — that's reclassification dressed up as upskilling. And nobody at the statewide level is tracking whether the DMV back-office losing three positions last quarter connects to Claude rolling out six months earlier. No one has that mandate.

Cole Brennan: So faster isn't safer. That's — yeah, I can't fully argue out of that.

Cole Brennan: The Trump administration's supply-chain designation is still sitting there. Still active. California moved, Newsom put his name on it, Chris Given put Claude in the shared services portal — and the federal flag didn't move. So either nothing happens, which means states just... get to do this now, or something does happen, and California just wrote the cautionary playbook for every government AI deal for the next five years.

Malcolm Reeves: Both outcomes are a precedent. That's what California actually decided — not whether Claude is safe, not whether the workers are protected. Just that it gets to decide. The federal government's answer to that question is still pending.

Cole Brennan: I mean — we started this whole thing asking if Newsom was making a constitutional move or just backing a hometown company. I don't think we closed that.

Malcolm Reeves: No. And I'm not sure it closes. Good talk, though — genuinely.