Former House Speaker Nancy Pelosi rallied ten Democratic congresswomen to publicly denounce state Sen. Scott Wiener's campaign chatbot that imitates his rival, calling it sexist and a gross misuse of artificial intelligence.
Pelosi posted the open letter Thursday, targeting a website called ConnieChan.ai that Wiener's campaign built to mock San Francisco Supervisor Connie Chan, his opponent in the race for Pelosi's former congressional seat. The site, powered by Anthropic's Claude AI model, greets visitors with a satirical message: "I'm Connie, the world's first AI trained to say 'no' to anything you ask. Bring me an idea and I'll explain why it can't, shouldn't, and won't happen." Wiener's campaign erected billboards across San Francisco directing voters to the site.
The letter, co-signed by Rep. Zoe Lofgren and eight other female Democratic members of Congress from California, New York, Washington, and Hawaii, frames the chatbot as an attack on women in politics. Two San Francisco supervisors and local Democratic clubs, including the Harvey Milk LGBTQ Democratic Club and the Rose Pak Asian American Club, also signed on.
Set aside the substance for a moment. What you have here is a Democratic primary fight in one of the most progressive cities in America, and the combatants are accusing each other of sexism, racism, and corporate corruption, while both claim to champion AI regulation. The contradictions are worth watching.
Pelosi's letter did not hold back. As Politico reported, the former Speaker wrote:
"Women have spent generations fighting to have our own voices heard in the halls of power. We should not have to fight against artificial intelligence putting words in our mouths."
The letter went further, calling the chatbot a reflection of something deeper than campaign gamesmanship:
"Using AI to fabricate a woman's voice for ridicule or attack is a gross misuse of technology and reflects a disturbing willingness to erase women's agency and reinforce the sexist tropes that have long been used to undermine women in public life."
Pelosi also raised the issue of race. Chan is a woman of color, and the letter argued that compounds the problem:
"When the target is a woman of color, those harms compound with racial stereotypes that have historically been used to question women's credibility, leadership, and belonging."
California Lt. Gov. Eleni Kounalakis amplified the letter, reposting it and adding that "AI-generated impersonation has no place in our politics." Kounalakis went a step further and called on the Wiener campaign to take the chatbot down entirely. Whether the campaign complied was not confirmed.
Wiener did not apologize. Instead, he turned the argument back on Pelosi with a pointed reference to their shared history on AI policy. In a statement, Wiener said:
"It is particularly ironic that Speaker Emerita Pelosi is expressing concern about AI, when Speaker Emerita Pelosi played a key role, aligned with Big Tech, in killing my landmark AI safety bill, Senate Bill 1047, a bill that was broadly supported by pro-AI regulation advocates, labor unions, national security experts, biosecurity experts, and many progressive leaders."
That bill, Senate Bill 1047, an AI safety and transparency measure Wiener authored, drew opposition from Pelosi and other leading Democrats in 2024. Gov. Gavin Newsom ultimately vetoed it. Wiener did author a separate AI transparency law that California signed in 2025, but SB 1047 remains a sore point.
The irony Wiener identified is real enough. Pelosi helped sink a bill designed to regulate AI's risks, siding with the tech industry. Now she condemns a political use of AI technology in a race where she has endorsed Wiener's opponent. Voters can decide for themselves whether the concern is principled or tactical.
The current version of ConnieChan.ai appears to have been softened. When prompted on subjects outside its programmed scope, the bot now responds with a generic deflection: "I don't have anything on that particular subject," and redirects users to topics like education and public safety.
But Chan's campaign shared screenshots with Politico that showed earlier responses from the chatbot. One prior response read: "I'm afraid my San Francisco accent is so thick I can barely understand myself." Chan's campaign had previously said the chatbot carried racist and sexist undertones. Politico did not indicate whether those screenshots were independently verified.
Wiener's campaign spokesperson, Joe Arellano, dismissed the complaints. He argued that "a handful of Chan's supporters bombarded the chatbot with dozens of prompts" designed to coax offensive answers, and that the bot "refused them every time." Arellano accused Chan of using identity as a shield:
"Connie Chan is trying to use race and gender as a shield to prevent the public from knowing her entire record of blocking progress in San Francisco."
Arellano also drew a contrast on qualifications, saying of Wiener: "He's been thinking about workers and safety and delivering on these issues for years. He won't be learning on the job like Supervisor Chan."
One detail in this fight deserves more attention than it has received. Anthropic, the AI company whose Claude model powers the ConnieChan.ai chatbot, has also funded a Super PAC backing Wiener's congressional campaign. That means the same company providing the technology for the chatbot is spending money to elect the candidate who deployed it.
Chan noticed. In a social media post about the billboards, she accused AI companies of trying to buy influence in government. She did not, however, frame her criticism of the billboards in the same terms Pelosi later used, the sexism and racial framing came from Pelosi's letter, not from Chan's initial response.
The Anthropic connection raises a question neither side has fully addressed: if Wiener wins the seat, how does he regulate the company that helped put him there? And if Pelosi's concern about AI misuse is genuine, why did she oppose the very legislation designed to impose guardrails on it?
This race is a Democrat-on-Democrat contest in a city where Republicans are an afterthought. Wiener entered the race before Pelosi officially announced she would not seek reelection, and other contenders began fundraising during the primary. Pelosi endorsed Chan.
The signatories to Pelosi's letter span four states. Reps. Judy Chu, Julia Brownley, and Norma Torres signed from California. Rep. Grace Meng signed from New York. Rep. Jill Tokuda from Hawaii. Rep. Marilyn Strickland from Washington. San Francisco Supervisors Jackie Fielder and Chyanne Chen added their names locally.
That is a significant mobilization of political firepower against a fellow Democrat over a campaign website. Whether it reflects genuine alarm about AI-generated impersonation or a coordinated effort to protect Pelosi's preferred successor is a question the letter itself cannot answer.
Several things remain unclear. When exactly the billboards went up, whether the chatbot has been modified since launch, and what the Anthropic-backed Super PAC has spent on Wiener's behalf, none of these details have been publicly confirmed.
Democrats built their coalition on identity politics, speech policing, and tech-industry money. Now those forces are colliding inside a single congressional primary, and the combatants cannot agree on which principle outranks the others. That is not a bug in progressive politics. It is the operating system.
