Politicians Wage War on Unflattering ChatGPT Descriptions

The New York Times reports that political campaigns are actively lobbying AI companies to change how their chatbots describe candidates — treating AI outputs the way a previous generation of campaigns treated Wikipedia: something to monitor, massage, and push back on when unflattering.

The concern is practical. When undecided voters ask Claude or ChatGPT about a candidate and get an incomplete or negatively framed answer, campaigns believe it can affect their numbers. According to the Times, some of these voters may trust AI responses more than traditional media — a claim that, if accurate, would make LLM representations of candidates a genuine electoral liability.

What's new is campaigns taking this seriously enough to pressure AI companies directly. That's a shift from asking platforms to moderate harmful content to asking them to change factual descriptions of public figures — a harder ask with murkier legal and ethical precedents. AI companies, unlike search engines, have direct editorial control over what their models say about anyone; campaigns are now trying to exert influence over that.

Read the full story at The New York Times