A group of current and former OpenAI employees have contributed more than $215,000 to a super PAC advocating for stricter regulations on frontier AI laboratories, creating an unusual internal rift at one of the world's most prominent artificial intelligence companies. The donations to Guardrails Alliance represent a direct counterweight to a pro-industry political spending group backed by OpenAI's own president and cofounder, Greg Brockman.
Rank-and-File Workers Push Back
Guardrails Alliance launched last month with $5 million in initial funding, positioning itself as a populist coalition of tech workers, labor unions, and other organizations. Seven current OpenAI employees and one former staffer have donated to the group, with some names shared exclusively with WIRED ahead of the super PAC's first quarterly filing with the Federal Election Commission on July 15.
The largest single contribution from an OpenAI employee came from Juan Felipe Cerón Uribe, a research engineer at the company since 2022, who gave $200,000. Cerón Uribe said he spent the last four years working on strategies to mitigate potential societal harms caused by AI.
In a statement, Cerón Uribe said he became concerned that his research would go to waste if it did not translate into guardrails holding private companies accountable for responsible AI development. He characterized tech billionaires such as Brockman as funding Leading the Future to keep AI unregulated, and described his decision to donate to Guardrails Alliance as easy.
Other OpenAI donors include safety researcher Gabriel Wu, who contributed $5,000, and AI alignment researchers Julie Steele and Jason Wolfe, who each gave $5,000. David Farhi, a former OpenAI research manager who departed last summer after seven years, donated $3,000 and will appear in the group's July filing.
Farhi said it was disappointing to see Leading the Future actively work against OpenAI's mission by aiming to shut down discussion around AI regulation before it could happen.
A David vs. Goliath Financial Gap
The contributions from OpenAI workers represent a fraction of Guardrails Alliance's broader goal to raise $15 million during this election cycle. They also pale in comparison to the $50 million commitment that Brockman and his wife, Anna, have made to Leading the Future.
Despite the disparity, Guardrails Alliance cofounder Shaunna Thomas, a longtime Democratic political organizer, said she is not worried about being outspent. Reaching $15 million would enable the group to follow Leading the Future into more political races, she told WIRED, but matching opponents dollar-for-dollar is not the strategy.
Thomas argued that when the public is made aware of what AI PACs are doing, people reject it. She said the group is leveraging existing public opinion, which is less expensive to do.
Thomas said she realized after Leading the Future's launch last year that politicians proposing new AI regulations would face a difficult conversation with a $100 million threat looming over them.
Leading the Future's Political Track Record
Leading the Future launched last summer with Brockman as one of its marquee supporters. The group has stated its goal is to oppose policies that stifle innovation and figures who support such an agenda.
Among the PAC's first political efforts was an attempt to derail the congressional campaign of Alex Bores, the author of New York's landmark AI safety law. Bores ultimately lost in a primary election last month. The group has since backed several pro-industry candidates across the country.
OpenAI's global affairs chief, Chris Lehane, previously told WIRED he helped establish Leading the Future and has advised Brockman on political giving, though he is not involved in day-to-day operations. An OpenAI spokesperson directed WIRED to a June blog post stating that Brockman's engagement with the PAC was in a personal capacity and that employees are free to participate in the political process personally.
Leading the Future spokesperson Jesse Hunt denied that the super PAC has tried to stifle public debate about AI, noting it has previously advocated for federal regulations on the technology. Hunt said the group has laid out a clear, positive, and proactive agenda and is proud of its track record supporting a diverse array of policymakers and candidates.
A Broader Counter-Movement
Guardrails Alliance is not the only group working against Leading the Future. Public First Action, a super PAC backed with $20 million from Anthropic, has committed to promoting AI safeguards and countering pro-AI groups in the 2026 elections. Thomas said Guardrails Alliance is distinctive because it represents a broad coalition without large corporate donors.
Both groups supported Bores in his primary race for New York's 12th Congressional District, a contest flooded with $27 million in spending from pro-industry and pro-safeguard organizations. Thomas said Guardrails Alliance is now evaluating support for other Democrats in the 2026 elections, including in California's 34th Congressional District.
Additional donors are expected in Guardrails Alliance's public filing, including former Andreessen Horowitz partner John O'Farrell, though the amount of his contribution was unclear. O'Farrell, who joined the venture capital firm in 2010 as its first outside partner and left earlier this year, criticized his former colleagues in a New York Times opinion essay for allegedly using Leading the Future to intimidate politicians engaging with questions about AI governance.
The donations from OpenAI's own workforce underscore a growing internal debate over the company's role in shaping AI policy, with some employees now using personal funds to directly oppose the political spending of their own leadership.
As the battle over AI regulation intensifies heading into the 2026 elections, the unusual sight of employees funding opposition to their own company's leadership highlights just how high the stakes have become. What do you think about tech workers taking a stand against their employers on policy issues? Share this article and join the conversation.
Source: Wired
