Former House Speaker Nancy Pelosi forced California state Sen. Scott Wiener to take down a parody AI chatbot targeting her endorsed congressional candidate, after Pelosi branded the stunt "sexist" and "racist" in a public broadside backed by fellow female officials.
Wiener launched the chatbot earlier in the week at the domain ConnieChan.ai, billing it as "the world's first AI trained to say no to anything you ask." The site was a jab at San Francisco city supervisor Connie Chan, whom critics have labeled a progressive NIMBY for opposing numerous housing proposals. By Friday, the chatbot was gone, pulled by Wiener himself after a 48-hour pressure campaign led by Pelosi turned the race's focus from housing policy to gender politics.
The sequence tells a straightforward story about how power works inside San Francisco's Democratic machine, and how a former Speaker still calls the shots even after leaving Congress.
On Thursday, Pelosi released a statement through her X account accusing Wiener of using artificial intelligence to impersonate a female candidate's voice. She framed the chatbot as an attack on women in politics broadly.
"Women have spent generations fighting to have our own voices heard in the halls of power. We should not have to now fight against artificial intelligence putting words in our mouths."
The statement went further, calling the chatbot a "gross misuse of technology" that "reflects a disturbing willingness to erase women's agency." Because Chan is a non-white woman, Pelosi's statement claimed, such harms "compound with racial stereotypes." Unnamed "numerous female elected officials" co-signed the message.
Pelosi endorsed Chan as her chosen successor for the congressional seat she held for decades. That endorsement was itself seen as a rebuke of Wiener, who had established himself as a frontrunner in the race. The two Democrats are scheduled to face each other in November.
So the chatbot fight is not really about artificial intelligence or sexism. It is about Pelosi defending her handpicked heir against a rival she does not want in her seat.
Wiener's initial response was sharp. He fired back on X, calling Pelosi's accusations "baseless" and her framing "classic deflection." He pointed out that his campaign had clearly labeled the chatbot as parody. And he landed a counter-punch with real policy weight: Pelosi, he noted, had opposed his first-in-the-nation AI safety law in California.
That is a genuine irony worth stating plainly. Pelosi invoked the dangers of AI to attack Wiener, but when Wiener tried to regulate AI through legislation, Pelosi stood against him. The accusation and the record point in opposite directions.
Yet by Friday, Wiener reversed course and pulled the chatbot down. His statement framed the retreat as a strategic pivot, not a concession.
"Our use of an AI chatbot clearly missed the mark. The conversation became about the tactic instead of the issues that actually matter to San Franciscans."
That language, "missed the mark," "the conversation became about the tactic", is the sound of a politician who knows he lost a news cycle and wants to move on. Wiener did not apologize for the chatbot's content. He apologized for its political cost.
Pelosi, who announced plans for a namesake institute at UC Berkeley after retiring from Congress, has shown no sign of loosening her grip on San Francisco's political landscape. Her intervention in this race is the latest evidence that retirement has not diminished her willingness to shape outcomes from behind the scenes.
The Chan campaign did not accept Wiener's concession gracefully. In a statement to The Post, her team escalated the language even after the chatbot came down.
"Scott Wiener can rightly take down his bigotry bot, but he can't erase the concerns it raised among San Franciscans about his character and judgment."
"Bigotry bot" is campaign rhetoric, not a factual description of a parody site that mocked a politician's voting record on housing. But the phrase tells you where Chan's team wants this fight to stay: on Wiener's character, not on Chan's record of opposing development.
And that is precisely what Wiener was trying to highlight before Pelosi intervened. Chan's critics describe her as a progressive who blocks housing proposals while San Francisco faces a well-documented affordability crisis. The chatbot, crude as it was, aimed at that record. Pelosi's response ensured the conversation never got there.
The relationship between Wiener and Pelosi has reportedly been tense, and this episode did nothing to ease it. Pelosi's endorsement of Chan over the primary frontrunner was already a signal. Her public attack on Wiener's campaign tactics was a louder one.
Former Speaker Pelosi has drawn scrutiny before for her combative approach to political rivals, and this episode fits the pattern. When a competitor threatens her preferred outcome, the response is swift and personal.
Step back from the personalities and the real question becomes clear: since when does a labeled parody chatbot qualify as sexism and racism?
Political parody is as old as American elections. Candidates mock each other's records constantly, in mailers, ads, debate stages, and now, apparently, through AI chatbots. Wiener built a site that satirized Chan's record of saying no to housing proposals. He labeled it as parody. He did not impersonate Chan or put fabricated policy positions in her mouth. He exaggerated her existing record for comic effect.
Pelosi's statement recast that as an act of technological misogyny, erasing a woman's agency, compounding racial stereotypes. Those are serious accusations. They deserve serious evidence. The material provides none beyond Pelosi's assertion.
What it does provide is context. Pelosi opposed Wiener's AI safety legislation. She endorsed his opponent. She has a tense relationship with him. And when he deployed a tactic that put her preferred candidate on defense over a weak policy record, Pelosi reached for the heaviest rhetorical weapons available in modern Democratic politics: sexism and racism.
Wiener, to his credit, identified this dynamic before he capitulated. "Classic deflection," he called it. He was right. Then he folded.
The race to fill Pelosi's seat has already drawn national attention and big money, and the November matchup between Wiener and Chan promises more of the same intra-party combat. Pelosi's financial activities have also kept her in the spotlight; she recently disclosed up to $6 million in stock options amid growing scrutiny of congressional trading.
The substance underneath this fight matters far more than the fight itself. San Francisco residents face some of the highest housing costs in the country. Chan's record on housing proposals is a legitimate campaign issue, arguably the central one in this race. Wiener wanted to make it the story. Pelosi made sure he couldn't.
Whether voters in November care more about a taken-down chatbot or about who will actually build housing in their city remains to be seen. But the episode reveals how Democratic power brokers use identity-based accusations to shut down policy debates they would rather not have.
Wiener labeled a parody site as parody, aimed it at a public official's public record, and pulled it down within days under political pressure. For that, he was accused of sexism, racism, and erasing women's agency. The charges came not from voters but from the most powerful Democrat in San Francisco, who happens to have a direct personal stake in the outcome.
When "sexist" and "racist" become all-purpose tools for protecting a favored candidate from legitimate criticism, the words lose their meaning, and the voters who actually face discrimination lose their vocabulary.