AI Giants Turn To George Soros-Funded Nonprofit For Election Information

AI companies are turning to a George Soros-funded organization to provide users with election information ahead of the pivotal midterm elections. 

Both OpenAI and Anthropic announced that they’re partnering with Democracy Works, a nonprofit that received money from Open Society, the liberal billionaire’s primary funding mechanism. The group also discloses funding from Pierre Omidyar, another liberal billionaire who has spent millions of dollars to fight President Donald Trump politically. 

Anthropic said the partnership is to help ensure that “generative AI users are safeguarded from mis-and-disinformation about voting and elections.” Open AI similarly said it will “partner with Democracy Works to display reliable information about voting and registration processes, including voting locations and other election logistics, when people ask about those topics.”

The group’s funding from Soros is consistent with its partisan activity. In recent years, the Brooklyn-based Democracy Works has championed Left-leaning policies and positions, including support for expanded mail-in voting. Its CEO, Luis Lozada, openly supported the antisemitic college protests, describing fines issued to Ivy League universities for failing to control the pro-Palestinian encampments as a “challenge to democracy.”

“Universities are punished for allowing students to organize protests on campus, individuals are removed from public spaces solely for expressing their opinions,” Lozada wrote.

In 2025, the universities settled with the federal government after protests on their campuses were deemed antisemitic. Columbia and Harvard University reached settlements with the Trump administration. Columbia’s fine was historically high — the federal government directed Columbia to pay $200 million to the government and $20 million to its Jewish employees “who were unlawfully targeted and harassed.” 

Democracy Works is a tool that labels itself as the “connective tissue” that helps “millions of voters navigate a complex system.” One of its products, TurboVote, aids in voter registration, sends election reminders, and requests mail-in ballots. In a section describing the legislative process to potential voters, it puts a peculiar focus on the legislative branch’s impeachment duties, one of the obvious duties that will be used by Democrats to get out the vote in the midterm elections. 

“Senators serve six-year terms and are responsible for crafting federal legislation, approving presidential nominations, and conducting impeachment trials,” the platform says, for example, in its description of what a U.S. senator does.

Screenshot of election information on Democracy Works platform

Anthropic and OpenAI could pull from this information and other Democracy Works resources to inform their users when they ask about voting. Anthropic, which runs Claude, told ABC News that a visual banner with information from Democracy Works will be shown to users who ask questions about the election. OpenAI’s ChatGPT will do similarly. Both Google and Meta told the outlet they would prioritize pulling information on elections only from official sources.

Both OpenAI and Anthropic have been repeatedly accused of leaning Left in political outputs. Trump’s former AI czar David Sacks has said that Anthropic became a landing spot for staffers in the previous administration.

“Guess where those Biden AI staffers went to work as soon as the admin was over,” he said. “Anthropic.”

Pentagon officials presented concerns about using Anthropic’s products for military technology, arguing the company could turn off the technology if the Pentagon went against guidelines.

“If they’re willing to impose red lines that limit lawful military use, what’s to stop them from enforcing a shutdown if they disagree with how we’re applying it,” an official said in March during a dispute between Anthropic and the Pentagon over the company’s military work.

In 2024, Sacks admitted that OpenAI’s ChatGPT is “programmed to be woke.” Billionaire Elon Musk, who also has his own AI company, attributed the politics being pushed by AI models to the people building them.

“A lot of the AIs that are being trained in the San Francisco Bay Area, they take on the philosophy of people around them,” Musk said. His AI company is headquartered in Memphis, Tennessee.