The Reason Why Glenn Beck Made a Weird AI George Washington

Glenn Beck wants you to know that the “final chapter” of his career involves interviewing a cartoon George Washington.

Last weekend, Beck dropped a preview of “George AI,” an AI chatbot he’s been building as part of a new platform called The Torch, launching January 5. The whole thing is exactly as weird as it sounds. Beck sits at a table across from a digital rendering of Washington who, for reasons that remain unclear, is wearing a black crewneck T-shirt and appears to have been hitting the gym. Washington sits in front of a podcast microphone like he’s about to drop some takes on Joe Rogan’s show.

Beck asks his AI Washington what he thinks America’s biggest problem is. Washington starts to answer in a formal tone about dangers to the republic, but Beck cuts him off. “Could you just dumb it down a little bit?” The AI explains it has 29 documented points to share. Beck interrupts again: “Just speak in today’s language.”

And so AI George Washington, supposedly drawing from the actual writings of the Founders, delivers a sermon that sounds exactly like ... Glenn Beck. America’s problems are “moral,” not political or economic. We’ve “drifted from the virtues that make liberty possible.” We need “discipline,” “faith,” and “character.” The fix won’t be found in Washington, D.C., but “in every home, every school, every heart.” As Washington says that the country has “grown skeptical of truth,” a sinister-looking image of someone holding a sign that says “Protect Trans Kids” flashes on the screen. Subtle!

Right Wing Watch put it best: “You will be shocked to learn that the AI George Washington created by Glenn Beck sounds exactly like what would happen if Glenn Beck built an AI George Washington to sound exactly like Glenn Beck.”

The whole thing is silly. Beck claims his personal collection of founding documents is the third-largest in the world, behind only those of the Library of Congress and National Archives. He says he’s built a proprietary AI system with an “electric fence” around it that prevents outside information from contaminating the pure wisdom of the Founders. He claims this means George AI won’t “hallucinate” like other chatbots because it’s based entirely on primary sources.

But then he tells it to stop talking like the Founders because it’s “tedious.”

At one point in the full video, Beck claims one of the Founders wrote “at a 70th grade level.” That’s not a typo. Seventieth grade. This is the guy building an AI system to teach Americans “honest history.”

Beck’s also been clear about what’s actually in this supposedly neutral database of founding documents. It’s not just letters and laws. It includes “thousands of sermons from the pulpits” of churches the Founders attended because, as Beck puts it, “almost everything in the Declaration and Bill of Rights came from the pulpits.” This is a specific ideological interpretation of American history baked right into what he’s presenting as objective, document-based AI.

And look, Glenn Beck is a crank. He’s always been a crank. He had his chalkboard connecting random things to George Soros on Fox News. He’s promoted countless conspiracy theories. The Torch is going to have a relatively small audience of Beck superfans (if they still exist), and the main takeaway from the preview is that it’s unintentionally funny.

I’ve been writing about this pattern for most of the year now. The playbook goes like this: Claim existing AI systems are biased against conservatives, build your own, insist yours is the “unbiased” one based on “real” information, and then produce outputs that conveniently align with right-wing politics.

Beck says George AI has an “electric fence” keeping out contamination. Elon Musk says Grok is “anti-woke.” Meta says Llama 4 is correcting a “left-leaning bias.” They all claim to be removing bias when they’re actually just shifting it rightward.

Back in May, Grok started randomly injecting references to “white genocide” in South Africa into completely unrelated conversations. Someone would ask about baseball stats or a cute cat video and get a lecture about white farmers being persecuted. This wasn’t a glitch. AI researcher Matthew Guzdial told 404 Media that xAI was likely “literally just taking whatever prompt people are sending to Grok and adding a bunch of text about ‘white genocide’ in South Africa in front of it.”

That’s crude propaganda, and it got caught because it was so obvious. But the more sophisticated version is what Meta did with Llama 4 or what the Trump administration is trying to do with federal AI contracts. They’re not clumsily inserting conspiracy theories. They’re redefining what counts as “neutral” and “objective” so that conservative viewpoints become the baseline.

The Trump administration’s executive order on AI is literally titled, “Preventing Woke AI in the Federal Government.” It defines “ideological neutrality” in a way that treats acknowledging concepts like systemic racism or unconscious bias as violations of that neutrality. Climate change becomes “radical climate dogma.” Antidiscrimination efforts become “ideological bias.” And companies that want federal contracts have to build AI systems that reflect this worldview.

Then there’s PragerU, which partnered with the Trump White House to create “The Founders Museum,” featuring AI Founding Fathers who literally quote modern conservative talking points. An AI John Adams says, “Facts don’t care about your feelings,” which is Ben Shapiro’s catchphrase. They took the second president of the United States and turned him into a ventriloquist dummy for a guy who wasn’t born for another two centuries.

Beck’s George AI fits right into this pattern. He’s doing what Musk is doing with billions of dollars, just on a smaller budget with worse lip-sync technology. The pitch is identical: Other AI systems are biased; ours is based on REAL documents—it’s OBJECTIVE. And then the output just happens to sound exactly like conservative talk radio.

This matters because AI systems are increasingly how people interact with information. When these tools are deliberately calibrated to reflect a specific political worldview while claiming to be neutral, they don’t just reflect existing beliefs. They shape new ones.

As I wrote back in April about Meta’s shift with Llama, this is what researchers call “computational propaganda” and “personalized persuasion.” It’s not traditional media bias. It’s AI systems that can systematically shape society’s views through interactions that feel personal and objective. OpenAI’s Sam Altman has warned that AI might become capable of “superhuman persuasion” well before it reaches general intelligence. When you combine that persuasive power with deliberate political bias dressed up as neutrality, you’ve got a recipe for something genuinely dangerous.

Beck’s fake Washington telling Americans their problems are “moral” rather than political isn’t going to reshape society on its own. But it’s a proof of concept. It shows how easy it is to build an AI system that validates whatever you already believe, present it as based on “primary sources” and “real documents,” and sell it to an audience primed to distrust other information sources.

The right-wing push to dominate the AI landscape is going to accelerate. We’ve already got Grok pushing conspiracy theories, Meta preemptively shifting its models rightward, the federal government mandating “ideological neutrality” defined by whoever’s in power, and PragerU putting words in the mouths of AI Founding Fathers. Glenn Beck’s buff George Washington in a black T-shirt is almost quaint by comparison.

But they’re all doing the same thing. They’re building machines that tell their audiences what they want to hear while claiming those machines are just telling the truth. And that’s a game where reality itself is what’s at stake.