Biden's White House spent months pressuring Facebook to silence Americans — and Mark Zuckerberg put that in writing to Congress.
Now the weapon they built to do it requires zero humans to pull the trigger.
And the people who built it never stopped looking for the next platform to plug it into.
Facebook's AI Fact-Checking System Destroyed a Journalist's Income Over a Quote He Never Said
Emmy Award-winning journalist John Stossel released a video on Facebook in September 2020 blaming western wildfires on poor forest management.
Facebook flagged it immediately: "False Information. Checked by independent fact-checkers."
The label linked to a Science Feedback fact-check built around one central quote: "Forest fires are caused by poor management. Not by climate change."
Stossel never said it.
His primary source, author Michael Shellenberger, never said it.
Nobody in the video said it.
Science Feedback's own site listed the quote's source as "Facebook users" — anonymous strangers with no connection to Stossel's reporting.
Stossel tracked down two of the fact-check's listed expert reviewers — Stefan Doerr of Swansea University and Zeke Hausfather of The Breakthrough Institute — and both admitted on camera they had never watched his video.
Both initially agreed the fact-check shouldn't have applied to his work.
Then Stossel told Science Feedback what the experts said.
Both men reversed course within days and defended the label.
Stossel's monthly Facebook income dropped 45 percent.
Meta's AI Censorship Machine Ran Automatically With No Human Involved
Political analyst Matthew Hoy documented the full operation in his new book Fact-Checking Frauds, and the mechanism he uncovered is the real story.
Meta's algorithm — not a human editor, not a human fact-checker — connected Stossel's video to an older Science Feedback report and applied the warning label with zero human review at any point in the chain.
Months before Stossel's video appeared, Meta was publicly touting a new system called SimSearchNet++ powered by artificial intelligence that could identify similar content across posts even when images and text had been slightly altered.
That same pattern-matching technology crawled Science Feedback's fact-checks, matched them to Stossel's video, and generated the "False Information" warning automatically.
City Journal contributor John Tierney got the same treatment.
His April 2021 piece on COVID mask mandates for children was flagged using a Science Feedback fact-check of a completely different article from a completely different outlet published three months earlier.
According to Meta's own numbers, that label triggers up to a 95 percent reduction in how many people see the flagged content.
Zuckerberg Confessed in Writing That Biden Officials Pressured Facebook to Censor Americans
In August 2024, Mark Zuckerberg wrote a letter to House Judiciary Committee Chairman Jim Jordan admitting that Biden White House officials had "repeatedly pressured" his teams "for months to censor certain COVID-19 content, including humor and satire."
Zuckerberg wrote that the officials "expressed a lot of frustration with our teams when we didn't agree."
He admitted it was wrong: "I believe the government pressure was wrong, and I regret that we were not more outspoken about it."
Jim Jordan's committee called it exactly what it was — proof that the Biden administration pressured Facebook to censor Americans, and Facebook complied.
Jordan had already published a December 2024 report titled Censorship's Next Frontier documenting Biden's push to extend that same model directly into every major tech company.
Jordan sent letters in March 2025 to 16 technology firms — including Meta, Google, Apple, OpenAI, and Microsoft — demanding all documents related to Biden administration pressure on their systems.
The pattern Jordan's committee uncovered is the same one that destroyed Stossel's income: government officials pressuring private platforms, platforms deploying automated systems to execute those directives at scale, conservatives bearing the cost.
The AI Censorship Infrastructure Is Still Active and Still Targeting Conservative Speech
Stossel's video label quietly disappeared sometime after he lost his defamation lawsuit against Meta and Science Feedback.
U.S. District Judge Virginia K. DeMarchi dismissed the case by ruling that Science Feedback's fact-check was "not capable of being proved true or false" — because fact-checking is just opinion dressed up as journalism.
Stossel and Tierney didn't even know their labels were gone until Hoy informed them while researching his book.
Nobody at Meta made an announcement.
The system censored them, quietly removed the labels after the lawsuit made them legally inconvenient, and nobody was ever held accountable for a dime of lost income.
Meta officially ended its IFCN partnership in the United States in January 2025, and Zuckerberg made a show of calling the old system censorship.
But the infrastructure — the automated matching systems, the activist fact-checking networks, the government pressure pipelines — doesn't disappear because Zuckerberg changed his tone.
Jordan's committee heard testimony from investigative journalist Lee Fang in February 2024 that these tools give "powerful entities the unprecedented ability to monitor, flag, and censor billions of individuals at a scale and scope never before conceivable."
What happened to Stossel wasn't a bug in the system.
It was a preview — and the people who built it are still funded, still organized, and still hunting for the next platform to wire it into.
Sources:
