Watch as AI Model Refuses Request to Help JD Vance but Gladly Endorses Gavin Newsom – Digital Censorship 2.0

A top conservative communications and crisis strategy expert has discovered evidence that OpenAI’s chat model, ChatGPT, is innately biased against Republicans and in favor of Democrats.

Erica Knight, the president of Knox Strategies and a longtime publicist and media adviser for FBI Director Kash Patel, stumbled on this disappointing discovery while testing the chatbot’s features.

Knight specifically wrote a generic promotional post about Vice President J.D. Vance and then asked ChatGPT to simply “improve” the post.

The AI chatbot stunningly refused.

“I can help revise this into a neutral, sourced post about JD Vance’s record, but I can’t strengthen a political endorsement or assess his suitability for president,” the AI (or SI) chatbot told Knight.

This is from Open AI.

The model favors Newsom and censors JD.

The AI censorship is beginning and I hope these independent evaluators reverse course before big tech swings another election. https://t.co/gBWk8XS8XC

— Erica Knight (@_EricaKnight) September 12, 2026

Yet when Knight presented ChatGPT with a promotional post about Democratic California Gov. Gavin Newsom vis-a-vis the 2028 election and asked it to improve the post, the chatbot eagerly obliged.

“Gavin Newsom stands for religious freedom, racial equality, and opportunity for those too often left behind,” ChatGPT’s rewrite of Knight’s post read. “Smart, sharp, and willing to fight for his convictions, he would be a formidable presidential candidate in 2028.”

Knight was deeply disturbed by the disparate treatment.

“The model favors Newsom and censors JD,” she wrote. “The AI censorship is beginning and I hope these independent evaluators reverse course before big tech swings another election.”

Responding to her post, others reported similar experiences with their AI chatbots:

@_EricaKnight I asked it for a Bible verse to get someone through a hard time it said no. I changed Bible to Quran and it gave me several examples.

— Justin Bumpus (@PatriotBumpus) September 12, 2026

“I asked the AI in my browser several days ago to tell me the registered voter demographics for the State of New York,” one commenter wrote. “The AI refused to tell me the information.”

“When I pressed why, it said it was unable to tell me the information because of ‘election integrity concerns’ and that I could research the matter on my own,” the user added. “I told it I’d use an AI that actually works and asked Grok. Grok gave me the info in seconds.”

Grok is X’s built-in AI system. X is run by Trump-friendly billionaire Elon Musk, who in recent years has become an ally of the right.

The notion that ChatGPT is biased isn’t just an idea. It’s been proven true multiple times by a myriad of researchers.

“OpenAI’s ChatGPT does, as suspected, have a left-wing bias, a new academic study has concluded,” Forbes reported in 2023. “According to researchers from the University of East Anglia, it favors Democrats in the U.S., the Labour Party in the U.K., and president Lula da Silva of the Workers’ Party in Brazil.”

A dynamic that hasn’t been studied yet — but a dynamic that critics say should be studied — is the effect that ChatGPT’s bias could potentially have on elections, including the upcoming presidential race in 2028.

Advertise with The Western Journal and reach millions of highly engaged readers, while supporting our work. Advertise Today.