Two things struck me reading Industrial Policy for the Intelligence Age, OpenAI's new policy blueprint, which Sam Altman is calling a "New Deal for the AI Era".
First, many of the policy reforms in the document poll extremely well, particularly on the left. Higher taxes on capital gains and corporate income. A nationally managed public wealth fund seeded by the AI companies themselves. Portable benefits untethered from employers. "Auto-triggered" unemployment benefits and cash assistance. A four-day workweek!
Public support for OpenAI's policy proposals — share who support each policy, by party
Democrats
All Americans
Republicans
Economic proposals
Raise capital gains taxes for millionaires
Four-day workweek at full pay
AI safety & transparency proposals
Government should regulate AI safety, even if it slows development
Mandatory safety testing before AI models are released
Independent expert safety evaluations of AI products
Sources: Pew Research Center (Jan. 2025) for corporate taxes; FinanceBuzz/Pollfish (Oct. 2024) for capital gains; Ipsos (2021) for four-day workweek; Gallup/Special Competitive Studies Project (Sept. 2025) and AI Policy Institute for AI safety proposals. Democrat figures include Democratic-leaning independents where noted.
Axios co-founder Mike Allen put it well when interviewing Sam Altman this week: "This reads like a Bernie Sanders fever dream."
The second thing that stood out: most of these policy proposals have no chance of becoming law any time soon, or even receiving a serious hearing in the current Congress.
In the Axios interview, Altman admitted as much: "I'm sure these ideas aren't all good," he said. "The goal here is really to start a conversation."
I agree with what many observers have noted: this is as much corporate PR as policy paper. It's an attempt to position OpenAI as the responsible actor in the room, win back left-leaning skeptics, and steer the conversation about AI guardrails without actually backing existing legislation.
To be clear, none of that is inherently malicious.
The real problem is the contradiction between OpenAI's rhetoric and its actions. The same executives who wrote the blueprint have spent the past year opposing politicians who support these ideas and lobbying for legislation that would undermine the worker protections the document claims to champion.
Here's a sampling of the rhetoric in this week's policy blueprint, and the actions of company executives and funders over the past year.
Rhetoric
"Invest in clear, real-time measurement of how AI is affecting work, wages, job quality, and sectoral dynamics, using public metrics such as unemployment rates and indicators of regional or industry-specific displacement."
Action
In the lead-up to the One Big Beautiful Bill, signed into law on July 4, 2025, OpenAI
lobbied for a 10-year moratorium on state AI regulation – including the reporting requirements and impact assessments that states would need to track displacement in the first place.
Rhetoric
"Allow workers to prioritize AI deployments that improve job quality by eliminating dangerous, repetitive, administrative, or exhausting tasks so employees can focus on higher-value work. At the same time, set clear limits on harmful uses of AI that could erode job quality by intensifying workloads, narrowing autonomy, or undermining fair scheduling and pay."
Action
In August 2025, OpenAI launched
Leading the Future, a super PAC backed by Andreessen Horowitz and Palantir whose explicit goal is to oppose AI regulation. OpenAI Co-founder Greg Brockman and his wife each contributed $25 million to the PAC (Brockman has admitted the goal of his political spending is to counter negative public opinion about AI, not advance worker protections). The PAC raised $125 million in 2025 and spent its first year targeting politicians who support worker protections and AI transparency. It
spent over a million dollars in attack ads against Alex Bores, a New York assemblymember running for Congress whose platform centers on protecting workers in the age of AI. Bores authored the
RAISE Act, which would require AI companies to publicly disclose safety protocols and publish transparency reports when deploying new models.
Rhetoric
"Policymakers could rebalance the tax base by increasing reliance on capital-based revenues—such as higher taxes on capital gains at the top, corporate income, or targeted measures on sustained AI-driven returns—and by exploring new approaches such as taxes related to automated labor."
Action
Marc Andreessen, co-founder of Andreessen Horowitz – the
largest institutional backer of OpenAI's super PAC –
called a Biden capital gains proposal "the final straw" that drove him to support Donald Trump in 2024. He has also argued that a universal basic income would turn people "into zoo animals to be farmed by the state" – a far cry from the auto-triggered safety nets and public wealth funds OpenAI's blueprint champions.
In other words, OpenAI's political machine is funded by people who have made clear, in both word and deed, that they have no interest in expanding the social safety net.
Anthropic – itself a private company with its own commercial interests – has at least published actual labor market data, funded independent research on AI's economic impact, and backed candidates who support enforceable transparency laws. That is what it looks like to take these questions seriously, even imperfectly.
OpenAI has done the opposite. It has published a wish list in an effort to position itself as a good-faith actor, while spending $125 million to undermine the very policies the blueprint claims to support. That is not just corporate PR. It is dishonest.
ZK
Zachey Kliger
Policy researcher and writer focused on American democracy and tech policy. He has worked with the American Academy of Arts & Sciences on the Our Common Purpose initiative.
