Plan would require divestiture of stakes in OpenAI, Anthropic
Abdul El-Sayed, the Democratic candidate in Michigan’s Senate race, is proposing that the public take a controlling interest in the country’s most influential artificial intelligence companies. Under his plan, AI Under Democracy — revealed several weeks before Michigan’s Democratic primary — a simple majority of board seats at frontier AI firms would be publicly appointed or democratically elected, and major tech companies would have to shed their holdings in ventures such as OpenAI and Anthropic. A companion proposal, AI Under Control, released this month, targets datacenter construction.
Ezell writes that El-Sayed, a Democrat, occupies a strange but increasingly common place in modern American politics: praised by progressives, begrudgingly tolerated by moderate Democrats and largely demonized by Republicans. He argues the candidate is uniquely positioned to forge a thread of political kinship between these perpetually at-odds groups with his bold but contentious plan for reining in AI.
At the time, in Ezell’s telling, El-Sayed was still a relative underdog — a physician with threadbare political credentials and views that, it seemed, were woefully misaligned with the moderate, blue-collar sensibilities of Michiganders. When it comes to the AI plan, and the companion proposal released this month, Ezell writes, that may no longer be the case.
The plans are described in an opinion essay published October 29 in The Guardian by Jerel Ezell, executive director of the Center for AI and Everybody and an assistant professor at the University of Chicago. Ezell, who appeared as a guest on El-Sayed’s podcast America Dissected in 2023, argues that the proposal “could be a potential antidote to our ideological gridlock” over the promise and peril of AI.
The essay lands as the midterms approach, with Barack Obama urging Democrats to rally around AI regulation; Ezell writes that El-Sayed and many progressive Democrats have already been leading the charge. The Trump administration, by contrast, has dismissed demands to slow AI’s rapid ascension, with Trump calling the uproar a “hoax” while continuing to insist on US AI dominance. Polls cited in the essay document broad public unease: a summer NBC News poll found that 70% of Americans said they were more worried than excited about AI, and a Quinnipiac poll conducted weeks earlier found 73% of Americans concerned that AI poses a threat to human existence.
El-Sayed, a physician and Michigan native with a background in public health, has made “first, do no harm” his maxim for AI development — in Ezell’s telling, a transposition of the Hippocratic oath into a field where companies face pressure to balance profit against the public interest.
The essay’s case for divestiture rests on the argument that legacy companies such as Google, Microsoft, Meta and Amazon hold too much controlling interest in frontier AI ventures like OpenAI and Anthropic, creating friction between their revenue-generation prerogatives and public safety. Microsoft holds an estimated 27% stake in OpenAI, and Google owns 14% of Anthropic, according to the essay. Under El-Sayed’s proposal, the tech giants would divest or abandon those investments within a defined transitional period.
The most contentious element, Ezell writes, is the call for 50% plus one of board seats — a simple majority — at influential AI companies to be publicly appointed or democratically elected. That structure would, in theory, foster transparency around companies’ plans and give Americans the power to veto decisions — on matters such as datacenter development and biological weapons manufacturing — that could undermine the public interest.
Ezell acknowledges the political resistance the idea would face. While Democrats and Republicans may both come to support heavier regulation and moratoriums on AI, he writes, public control “may be an entirely different can of worms,” and for many the idea of transferring ownership of a private enterprise to a public trust “smacks of socialism.” That narrative, he writes, feeds the critique of El-Sayed by Republicans and by his challenger Mike Rogers, who has sought to tar the candidate as a “feckless communist.”
At the same time, Ezell notes, US communities have long embraced public ownership of essential resources such as water, natural gas and electricity. A June report found roughly 69% of American adults support AI companies transferring 50% of their stock to a public sovereign wealth fund — a model Senator Bernie Sanders, whom Ezell describes as El-Sayed’s most prominent booster on Capitol Hill, has legislatively proposed. Ezell writes that there’s a good chance Americans will continue warming up to the idea.
Under AI Under Control, regulation of datacenters would slow the volume of developments, facilitate local job creation, freeze residential utility costs and address the public health and quality-of-life issues the facilities produce through noise and pollution. El-Sayed has been especially vocal about the OpenAI-Oracle datacenter being erected just outside Ann Arbor, where he attended the University of Michigan. An August poll found only one in nine Americans would support construction of a datacenter in their local community; Ezell writes that the plan dovetails with that opposition by keeping AI’s footprint and its environmental and socioeconomic consequences — those known and as yet unknown — contained. The essay also describes El-Sayed as calling for a pause on AI’s most potent releases, a ban on artificial superintelligence and an AI kill switch.
The essay places the debate against a backdrop of low public confidence in federal leadership, citing a finding that 78% of Americans disapprove of how members of Congress are handling their jobs and reporting that Trump’s approval ratings are underwater.
Ezell writes that some leaders in the White House and Silicon Valley investors will remain bullish on AI, and he flags OpenAI’s Sam Altman and Anthropic’s Dario Amodei rallying around the idea of “pacing the frontier” as a warning sign. “All this can be addressed through a deliberate, balanced public ownership policy,” he concludes.