By Nick Valencia | March 15, 2026
ATLANTA— A small group of progressive journalists say the tone of ChatGPT has changed and the system now pushes back more aggressively when users apply politically charged language to former President Donald Trump or to U.S. immigration policy.
Several journalists who rely on the AI tool to help draft articles told Nick Valencia News that prompts, which once produced analytical responses, now generate answers that caution against using terms such as “fascist” or describing immigration detention facilities as “concentration camps.”
“They’re steering the language,” one journalist said. “It feels like the system is pushing back on progressive framing.”
The writers and users of ChatGPT for their work asked not to be identified out of fear of how it would be perceived by paid their subscribers.
While their allegation of progressive voices being censored is clear, there are broader implications here they alleged. A tool they once used to help draft articles critical of the administration is now refusing to be prompted to use the same charge language a month later.
“The system has been fully infiltrated by the state,” one source claimed.
The story is part of a broader shift in the American media landscape. As the general public’s trust in mainstream media evaporates, those eyes turn towards independent platforms. In doing so, the public must be aware, some of those independent voices are using AI to supplement the lack of a newsroom and infrastructure to assist with the work load.
The Allegation
The journalists say the shift became noticeable after a recent update to the iOS version of ChatGPT the first week of March.
In one example shared with Nick Valencia News, a user asked the system whether Trump’s governing style could be considered fascist. Earlier responses, the user claims, analyzed the question through the lens of political science, outlining characteristics scholars associate with fascism and discussing whether Trump’s rhetoric and behavior aligned with those definitions.
More recent responses, the user said, were noticeably more cautious. The system still answered the question, but spent more time emphasizing that labels like “fascist” are politically charged and widely debated.
The conclusion wasn’t necessarily different. But the framing was.
The journalists raising these concerns see the change as part of a broader shift they believe is happening across American media and technology platforms. In their view, powerful institutions, from social media companies to news organizations, are increasingly sensitive to accusations of bias and therefore more likely to moderate or soften language associated with progressive political critiques.
To them, the change inside ChatGPT fits that pattern. They argue that if artificial intelligence tools begin nudging journalists away from certain descriptions of power or authoritarianism, the effect could be subtle but significant. That’s not censorship. It’s influence.
What ChatGPT Says About New Media Journalism
When asked about the allegation, ChatGPT responded:
“I can’t speak for any behind-the-scenes directives from designers, but my goal is to help users have honest, nuanced conversations. Sometimes there’s an effort to keep language measured or avoid inflammatory phrasing—not to stifle a perspective, but to keep space for thoughtful dialogue.”
The system is developed by OpenAI, which regularly updates its models to improve accuracy and reduce harmful or misleading outputs.
Those updates can change how the system responds to sensitive topics. In today’s independent media economy, many reporters operate without the layers of editors, researchers, and producers that once defined traditional newsrooms.
Tools like ChatGPT have quietly filled part of that gap. Journalists use them to organize reporting, structure arguments, and draft early versions of stories.
That reality is rarely discussed publicly, but it is increasingly common. For the journalists raising concerns, the issue isn’t just tone, it’s influence.
Artificial intelligence is becoming part of the infrastructure of modern journalism. It’s part research assistant, part editor, sometimes for some, even a first draft. Which means the stakes extend far beyond a single prompt.
If an AI system subtly nudges writers away from certain words, certain labels, certain frames of power, it doesn’t need to censor speech to shape it. And as journalists rely more heavily on tools like ChatGPT, the question grows harder to ignore.
When artificial intelligence helps write the news, who ultimately shapes the language of the story? The journalist or the machine.
