Anthropic is building a dedicated presidential engagement program for the 2028 US election: an in-house team that will work directly with presidential candidates from both parties on AI issues, advise company leadership on political strategy, and run Anthropic's political giving. Fortune reported the plan on October 8, 2026, based on a job description Anthropic posted for a Political Programs lead.
It does not change Claude as a product. It does show how much AI companies now treat Washington as part of the stack, which is the same theme we have tracked across the FTC probe of OpenAI and Anthropic, the White House super intelligence task force, and the White House accord.
TL;DR
| Question | Answer |
|---|---|
| What is it? | An in-house program to engage 2028 presidential candidates of both parties on AI policy |
| Who leads it? | A Political Programs lead Anthropic is hiring, salary range reported at $295,000 to $345,000 |
| What will they do? | Map candidate fields, offer AI policy education "on equal terms," prep Anthropic executives, plan for platform, convention and transition periods |
| Does it include political money? | Yes, the role runs "the day-to-day of Anthropic's political giving" |
| Who is critical? | The Center for Political Accountability and Sen. Alex Padilla (D-Calif.), per Fortune |
| Has Anthropic commented? | Fortune says Anthropic did not immediately respond |
What the job description says
The posting says the hire will "build and run Anthropic's presidential engagement program for the 2028 cycle." The duties Fortune quotes include mapping "the prospective fields in both parties," offering "AI policy education to candidates and their advisors on equal terms," preparing Anthropic's executives for those conversations, and planning "ahead for the platform, convention, and transition periods."
The posting's own framing is blunt: "Decisions made over the next two years will shape the future enormously. We want every serious candidate and elected official, in both parties, to understand the technology and the case for governing it well." It also describes Anthropic as "equal parts research lab, policy think tank, and technology company."
The role goes beyond candidate briefings. It includes grassroots organizing, running the day-to-day of political giving, and helping employees who choose to take part in politics do so "in an informed way." The candidate profile is a political operative: 12 or more years in campaigns, party committees or political organizations, senior national experience, and a bipartisan track record.
Why now: the context around it
Fortune places the move before the November 2026 midterms and an expected Anthropic IPO that it says could value the company at as much as $2 trillion. It argues the plan "underscores the extent to which government policy, and the support of the White House in particular, have become critical to the AI industry."
Anthropic's relationship with the current administration has been rough. Fortune reports that the Department of Defense earlier this year designated Anthropic a "supply chain risk" after a clash over military uses of its technology. CEO Dario Amodei has called for slowing AI development to shore up safeguards and has argued for regulation, a view Fortune says President Trump rejects out of concern about falling behind China. Even so, Amodei attended a White House event last week with other AI CEOs and signed a voluntary accord committing to self-regulation.
Anthropic's political footprint was already growing. Fortune notes that it set up its first employee-funded PAC, AnthroPAC, in April, with a legal cap of $5,000 per year per employee, and that Amodei gave $1 million in July to a pro-regulation super PAC called Public First, citing Politico. OpenAI, by contrast, says it has no employee-funded PACs, a policy it highlighted in a June 2026 blog post, according to Fortune, while its president Greg Brockman gave $25 million to a rival pro-AI super PAC, Leading the Future, before canceling a second planned $25 million after employee backlash.
For a longer look at how Anthropic's stance on rules has played out in public arguments with rivals, see our post on Altman and Anthropic on regulation.
What the experts say
Fortune collected three reactions, and they disagree.
- Bruce Freed, president of the non-partisan Center for Political Accountability, said companies usually reach campaigns through government-affairs staff, but he had not seen "such a formalized engagement through a presidential election engagement committee." His reading: "an AI company effort aimed at shaping or staving off government regulation or oversight."
- Katie Harbath, CEO of Anchor Change and a former Facebook public policy director, called it the start of "developing relationships with whomever will be in the White House in 2029." She said the role did not look unusual compared with what companies like Facebook did, with one exception: it includes running PAC money, which was usually separate. She also warned it may be hard for Anthropic to earn trust across both parties.
- Sen. Alex Padilla (D-Calif.) said the posting is a reminder that frontier AI companies "are also building political operations alongside their products." He added that "education and influence are not the same thing" and urged Anthropic to share more detail.
Cornell's Sarah Kreps told Fortune she expects AI to be a bigger issue in 2028 than it is now, shifting from the abstract to jobs, data centers and privacy.
Education or influence? How to read the program
The central argument in the coverage is where policy education ends and lobbying begins. There is no legal line that makes briefing a candidate neutral. A briefing decides which risks are framed as urgent, which technical claims are treated as settled, and which regulatory models sound reasonable. A company that sends a former campaign operator to explain "the case for governing it well" is making a case, however sincerely, and a candidate hears it as one voice among many.
That is also why the phrase "on equal terms" matters. If Anthropic truly offers the same access to every serious candidate in both parties, the program is closer to a standing public-affairs function than a partisan operation. If access is uneven, or if briefings are followed by donations, the optics change quickly. Padilla's request that Anthropic share more detail is, in effect, a request for that evidence: who is briefed, how often, with what materials, and whether the political giving arm and the education arm are kept apart.
Harbath's observation about PAC money is the most concrete structural point in the piece. In many large companies the people who brief policymakers and the people who direct political contributions are separate teams, partly so that substance and money do not look linked. Putting both under one lead is a design choice, and one the company may need to defend.
Questions readers are asking
Is this legal and routine? Companies lobbying and giving through PACs is established practice, and Fortune says so directly. What is unusual, per Freed, is the formal presidential-cycle framing.
Is Anthropic the only lab doing this? No. Fortune points to OpenAI hiring a congressional lead and to large donations from OpenAI leaders to outside groups. The difference is mostly structure and openness: Anthropic is advertising a candidate-facing program publicly, while OpenAI's influence, in Fortune's phrase, is subtler.
Will it change how Claude behaves? There is no indication of that. Anthropic's published positions on election integrity and political even-handedness for its models are a separate matter from corporate political strategy, and nothing in the job description touches model behavior.
Why start four years out? The posting itself says decisions over the next two years will shape the future "enormously," and primaries begin well before the general election. Relationships built early outlast any single campaign.
What it means for developers and businesses
First, regulation will likely be settled by the people who win 2028, and companies are positioning now. The rules at stake, such as export controls, safety testing, procurement terms and liability, shape what models you can use and what they cost. Our earlier coverage of the US government ban on Fable 5 and Mythos 5 showed how fast a policy decision can change access for builders.
Second, vendor risk now includes political risk. If you build on Claude, watch the Pentagon designation and the ongoing federal fights; our reporting on Meta and Microsoft cutting internal Claude use illustrates how customer decisions can move with the political weather.
Third, expect rivals to match. Fortune notes OpenAI is hiring a US Congressional Lead to advance its legislative priorities. A lab with no candidate-facing program will be at a disadvantage in 2028 conversations.
What we do not know
- Whether Anthropic will publish how the program works, including its funding and candidate selection.
- How "equal terms" will be audited in practice.
- Who gets the job, and whether the program will brief only major-party candidates.
- Anthropic's response to the criticism. Fortune says the company did not immediately reply.
This post relies on Fortune's reporting of the job description; we could not retrieve the posting itself, so quotes are as Fortune reproduced them.
Related reading
- FTC probe of OpenAI and Anthropic over AI safety
- White House super intelligence task force and Bessent on slowing down
- White House accord on super intelligence and frontier responsibilities
- Altman: accept some bad things, Anthropic and regulation
- US government bans Fable 5 and Mythos 5: Anthropic and export control
- Meta and Microsoft cut internal Claude use
- Official reporting: Fortune, October 8, 2026
Details reflect Fortune's October 8, 2026 report and may change as Anthropic comments.