AI safety groups aren’t buying OpenAI CEO Sam Altman’s shifting stance on political spending in the midterms.

They’re wary of his concession that campaign cash from industry-aligned groups, like the OpenAI-aligned Leading the Future super PAC, may have inadvertently backfired. Instead, they’re pushing for stronger commitments to downsize the flood of money — and for candidates to reject any support from pro-AI groups.

“I won’t believe that they are actually changing their strategy, or their goals I should say, unless and until they do,” said Democratic New York assemblymember Alex Bores, who emerged as a key voice among AI regulation advocates after Leading the Future and other groups spent $8 million to tank his House campaign. He launched a $30 million effort to unify Democrats on AI policy last month.

Altman’s stark acknowledgement to POLITICO follows a New York Times report that revealed OpenAI President Greg Brockman backed out of injecting more cash into Leading the Future earlier this summer, calling the super PAC’s efforts a “distraction” from the company’s mission. The pause is notable as Americans grow increasingly anxious about the risks the technology poses, and lawmakers scramble to figure out how to approach the emerging tech.

Still, some critics are not taking Altman’s concession or Brockman’s pullback as a sign of a fundamental change in the industry’s campaign strategy — at least not yet.

“What I read it as is they are no longer — or Greg Brockman in particular is not — going to fund Leading the Future,” said Bores. “Does that mean that the industry is not going to spend hundreds of millions of dollars trying to influence policy and get benefits on their way? No.”

Bores insisted that candidates should think twice before accepting financial support from Leading the Future specifically: “An elected official of either party taking money from Leading the Future should be viewed with incredible suspect.”

Guardrails Alliance political director Charlie Blaettler agreed, saying OpenAI leadership deciding to distance themselves from Leading the Future should be the final straw for Democrats backed by the group.

“If it’s too toxic for Sam Altman and Greg Brockman, why are Democrats still taking their money?” he said.

Leading the Future and rival super PAC network Public First — which advocates for AI regulation and is largely funded by OpenAI competitor Anthropic — together spent over $70 million through super PACs this cycle, according to FEC data.

Other pro-regulation advocates have jumped in with their own political spending, with groups like the Guardrails Alliance — a nonprofit and super PAC that launched in June — raising $5 million and spending hundreds of thousands of dollars to support Bores. Some of Guardrails Alliance’s money came from OpenAI employees pushing back against Leading the Future, which is funded in large part by OpenAI president Brockman, his wife Anna, early OpenAI investors Marc Andreessen and Ben Horowitz, and their venture capital firm a16z. They have contributed a combined $75 million to Leading the Future.

Leading the Future and a16z declined to comment.

Some Democrats who were backed in their primaries by Leading the Future — like Reps. Ritchie Torres and Joe Morelle of New York and Jared Moskowitz of Florida — declined to say whether they would disavow the Leading the Future cash that boosted them in their races.

“An outside AI PAC spent money in the 4th district during the primary, but they didn’t donate directly to my campaign,” Rep. Val Hoyle (D-Ore.), who was backed by a nearly $300,000 investment from Leading the Future, said in a statement to POLITICO.

She did not address whether she would disavow the group and its spending: “I’ve been working with labor leaders and advocates to put workers first when it comes to AI. We need clear federal guardrails to protect workers, safeguard our planet, prevent utility rate spikes and give consumers a voice.”

Even if candidates decide to further distance themselves from the industry’s spending, some pro-guardrails advocates said the onus still lies with figures like Altman to put a stop to — or significantly decrease — the influx of money altogether.

“[Altman was] basically saying, ‘These are the rules of the game, and I can’t really change them, and it’s not in my job description,’” said Brendan Steinhauser, who leads AI safety group The Alliance for Secure AI Action. “Of course he’s saying that, but they could do a unilateral disarmament or encourage the other labs to stop giving, and maybe it’d be just like pacing the frontier — maybe people could pace political giving.”

Former Democratic Rep. Brad Carson, who co-founded Public First, said he would be open to engaging in talks with Leading the Future about pacing or scaling back the industry’s political spending — but only if so-called AI “accelerationists” are serious about backing off. He called out Andreessen and Horowitz who have emerged as the biggest known spenders in this year’s midterm elections and were early investors in OpenAI.

“As long as these accelerationists have a desire to spend tens and tens, if not hundreds of millions of dollars, there is probably a need for a Public First,” said Carson, who also leads the nonprofit group Americans for Responsible Innovation. “So even if Brockman drops out, Andreessen and Horowitz can just drop another 50 million if they felt like it, and I wouldn’t want to be disarmed against that.”

Brendan Bordelon contributed to this report.