Documents revealed by Twitter’s new owner, tech billionaire Elon Musk, show the social media company has been intertwined with a government–private censorship apparatus.
Twitter suppressed or removed content on various subjects, including irregularities in the 2020 elections, mail-in voting issues, and various aspects of the COVID-19 pandemic. The company was under government pressure to purge such content and its purveyors from the platform, though most of the time it was cooperating with the censorship requests willingly, the documents indicate.
Musk took over Twitter in October, taking the company private. He then fired around half of the staff and much of the upper management, vowing to take Twitter in a new direction. The “#TwitterFiles” releases have been part of his promised focus on transparency for the company.
He allowed several independent journalists to submit search queries that were then used by Twitter staff to search through the company’s internal documents, sometimes under the condition that the resulting stories would be first published on the platform itself.
The two journalists primarily responsible for the releases have been journalists Matt Taibbi, a former contributing editor for Rolling Stone magazine, and Bari Weiss, a former editor at both The New York Times and The Wall Street Journal. Both are liberals who have expressed disillusionment with the more extreme currents of progressivism and neoliberalism.
Others involved in the releases have been independent journalists Lee Fang and David Zweig, former New York Times reporter Alex Berenson, and author and environmentalist Michael Shellenberger.
The journalists have only released a fraction of the documents they reviewed. They’ve also redacted the names of employees involved, aside from some high-level executives.
The documents show that the FBI and other state, local, and federal agencies have been scrutinizing the political speech of Americans on a significant scale, and trying to get lawful speech suppressed or removed online. Many conservative and traditionally liberal commentators have deemed that a violation of the First Amendment.
Twitter, a major hub of political speech, has been among the main targets of censorship. Many news stories have broken on Twitter in recent years, and a significant portion of the nation’s political debate takes place on the platform, as it allows an efficient way for direct and public interaction between users, from the most prominent to the least.
Twitter resisted some censorship requests, but there was little sign the company did so as a matter of principle. Rather, executives sometimes couldn’t find a policy they could use as a justification. Prior Twitter CEO Jack Dorsey was under pressure from his lieutenants to expand the policies to allow more thorough censorship, the documents show.
“We got Jack on board with implementing this for civic integrity in the near term, but we’re going to need to make a more robust case to get this into our repertoire of policy remediations—especially for other policy domains.”
In many cases, Twitter leaders de facto allowed the government to silence its critics on the platform.
Many censorship requests came in with an imperious attitude, particularly those from the Biden White House, but also some from the office of Rep. Adam Schiff (D-Calif.), who at the time headed the powerful House Intelligence Committee.
Schiff’s office accused Sperry of harassment and promoting “false QAnon conspiracies.”
Schiff’s demands were apparently a response to Sperry’s articles that speculated on the identity of the White House whistleblower that alleged a “quid pro quo” between President Donald Trump and Ukrainian President Volodymyr Zelenskyy.
Sperry reported, using anonymous sources, that the whistleblower was likely then-CIA analyst Eric Ciaramella, who was overheard talking in the White House with Sean Misko, a holdover staffer from the Obama administration. Misko later joined Schiff’s committee.
Under PressureThe many censorship requests Twitter received via the FBI were phrased as merely bringing information to its attention, leaving it up to the company to decide what to do with them. But Twitter executives clearly felt compelled to accommodate these requests, even in cases where they internally struggled to justify doing so, the documents show.
The government pressure took several forms. The FBI would follow up on its requests, and if they weren’t fulfilled, Twitter had to explain itself to the bureau. If Twitter’s position on an issue differed from the one expected by the government, company executives would be questioned and made aware that the bureau, and even the broader intelligence community, wasn’t happy. That would send the executives into triage mode, rushing to salvage the relationship, which they apparently considered essential.
Corporate media served as another pressure point. If Twitter wouldn’t do what it was told fast enough, the media would be provided with information portraying Twitter as ignoring some problem of paramount importance, such as possible foreign influence operations on its platform.
One censorship request, for instance, targeted an account allegedly run by Russian intelligence, though Twitter wasn’t given any evidence of it.
“Due to a lack of technical evidence on our end, I’ve generally left it be, waiting for more evidence,” said one Twitter executive, who Taibbi says previously worked for the CIA.
“Our window on that is closing, given that government partners are becoming more aggressive on attribution and reporting on it.”
The internal email suggests that Twitter, despite having no concrete evidence to back it, wouldn’t dare to disobey the request because of the media fallout of the government publicly labeling the account as being run by Russian intelligence.
Congress was perhaps the heaviest sword of Damocles hanging over Twitter’s head. Lawmakers could not only spur negative media coverage, but also tie up the company in hearings and investigations, or even introduce legislation that could hurt Twitter’s bottom line.
In the meantime, Twitter managers were convinced that lawmakers were leaking information that Twitter had provided to them and seeding negative news stories, even as the company was trying to placate them with increasingly stringent actions toward actual and alleged Russia-linked accounts.
Even though the FBI was officially only alerting Twitter to the activities of malign foreign actors, many of the censorship requests were simply lists of accounts with little to no evidence of malign foreign links. At times, Twitter tried to ask for more information, noting that it couldn’t find any evidence on its end, but often it simply complied. It was impossible for Twitter to do its due diligence on each request—there were simply too many, according to Taibbi.
“Without further explanation, Twitter would be forwarded an excel doc,” Taibbi said.
Censorship requests were lopsided against the political right. Some researchers said that the right was much more involved in spreading misinformation, but the documents indicate that the censorship wasn’t so much a matter of a right–left dichotomy, but rather a pro- and anti-establishment one. Even some left-leaning accounts were targeted if they strayed too far from the official government narrative.
Moreover, the right didn’t appear too keen on demanding censorship to begin with. Taibbi couldn’t find a single censorship request from the Trump campaign, Trump White House, or even any Republican, though he was told there were some.
On the other hand, there seemed to be no appetite across the board for targeting misinformation coming from the establishment itself.
Hunter Biden’s LaptopTwitter’s suppression of the 2020 New York Post exposé on Hunter Biden, son of then-candidate Joe Biden, was dissected in the Twitter release in particular detail. Apparently, some Twitter executives, particularly Roth, head of trust and safety, were regularly invited to meetings with the FBI and other intelligence agencies to receive briefings on the online activities of foreign regimes. In the several months prior to the 2020 election, Roth had been conditioned to expect a “hack-and-leak” Russian operation, possibly in October and involving Hunter Biden.
The FBI alleged there was some evidence of a Russian influence operation related to Hunter Biden’s dealings in Ukraine. But the bureau was also aware that Hunter Biden left his laptop with a trove of explosive information in a Delaware computer repair shop and that a copy of it was handed to Trump’s then-lawyer, former New York Mayor Rudy Giuliani. The FBI picked up the laptop from the repair shop in December 2019 and had Giuliani under surveillance in August 2020, when the repairman gave him the copy. As the FBI knew, the laptop information was neither hacked nor a figment of a Russian plot.
When the Post broke the story, Twitter executives were left with no doubt it was exactly what the FBI had been warning about.
Roth noted that the story didn’t actually violate any of Twitter’s rules. Nevertheless, it was marked “unsafe” and blocked on the platform under its policy against hacked materials, despite there being no evidence the materials were hacked.
Twitter’s then-Deputy General Counsel James Baker backed the censorship move, saying it was “reasonable” to “assume” the Hunter Biden information had been obtained through hacking.
Baker was the general counsel of the FBI until May 2018. He joined Twitter in June 2020. At the FBI, Baker was closely involved in the Russia investigation scandal, in which the FBI embroiled the Trump campaign and later the Trump administration in exhaustive investigations based on paper-thin and fabricated allegations that Trump colluded with Russia to sway the 2016 election. The allegations were produced by operatives funded by the campaign of Trump’s opponent, former Secretary of State Hillary Clinton.
ShadowbanningTwitter has long denied the practice of shadowbanning—suppressing the reach of an account without informing the user. The denial, however, specifically defined shadowbanning as making the person’s content invisible to others. What people have been complaining about is that Twitter seemingly suppresses how many people see their content, without making it invisible altogether. Internal materials show that Twitter has been doing that a lot, in fact.
Among those whose accounts were surreptitiously throttled was Jay Bhattacharya, Stanford University professor of medicine and one of the early critics of the COVID-19 lockdowns.
COVID-19Twitter has extensively suppressed information regarding the COVID-19 pandemic. Anything about the origins of the virus, its treatment, the vaccines developed for it, and public policies to mitigate its spread had to align with the official position of the federal government, as promulgated by the Centers for Disease Control and Prevention (CDC).
Zweig said he “found countless instances of tweets labeled as ‘misleading’ or taken down entirely, sometimes triggering account suspensions, simply because they veered from CDC guidance or differed from establishment views.”
Twitter user @KelleyKga, a self-described fact-checker, criticized a tweet that falsely claimed that COVID-19 was the leading cause of death by disease in children. @KelleyKga pointed out that such a claim would require cherry-picking data, backing that argument with data from the CDC. @KelleyKga’s criticism, however, was labeled as “misleading” and suppressed. On the other hand, the tweet that contained the false claim was not suppressed.
All physician Euzebiusz Jamrozik did was write on Twitter an accurate summary of study results on COVID-19 vaccine side effects. The tweet was labeled “misleading” and suppressed.
Sometimes, it appears, Twitter suppressed the information on its own, but many of the COVID-19-related requests came from the government and even directly from the Biden White House, internal files show.
In one email, White House Digital Director Rob Flaherty accused Twitter of “bending over backwards” to resist one of his censorship requests, calling it “total Calvinball”—a game where rules are made up along the way. The email, which wasn’t part of the Twitter files, came out during an ongoing lawsuit against the Biden administration filed by the attorneys general of Missouri and Louisiana.
Another White House staffer wanted Twitter to censor a tweet by Robert Kennedy, Jr., a long-time critic of vaccination. The staffer wondered whether Twitter could “get moving on the process for having it removed ASAP.”
“And then if we can keep an eye out for tweets that fall in this same genre that would be great,” he said in the Jan. 23, 2021, email.
The administration wasn’t always trying to get such content removed. People who merely expressed “hesitancy” about the vaccines were supposed to only have their content suppressed from reaching any significant audience, the documents indicate.
The Biden administration had a lot at stake, as the vaccine rollout was one of its first and most high-profile tasks. There were other stakeholders as well.
Several censorship requests came from Scott Gottlieb, board member and head of the regulatory and compliance committee at Pfizer, the pharmaceutical giant that made the most popular COVID-19 vaccine and raked in tens of billions of dollars on sales of it over the past two years.
Gottlieb sent Twitter at least three requests. One targeted a doctor who argued on the platform that naturally acquired immunity to COVID-19 was superior to vaccination. Twitter suppressed the tweet, even though the doctor was correct.
Another request targeted author Justin Hart, who argued on Twitter against school closures, pointing out that COVID-19 fatalities among children were extremely rare. Gottlieb sent the request shortly before Pfizer received approval for the use of its vaccine on children. Twitter didn’t comply with the request.
Yet another request targeted former NY Times reporter Berenson. Gottlieb claimed that Berenson’s criticism of Dr. Anthony Fauci, head of COVID-19 response in the Biden administration, was causing threats of physical violence toward Fauci. Twitter suspended Berenson’s account shortly after.
Gottlieb sent his requests to the same Twitter official who served as a contact person for censorship requests coming from the White House.
Trump DeplatformingTrump was particularly effective on Twitter. His soundbites, honed over decades of dealing with the New York press, played well on the brevity-oriented Twitter, earning the president some 90 million followers and lending him the power to bypass media filters and instantly grab national attention. Trump’s Twitter presidency, however, brewed scorn inside the Beltway, especially among the foreign policy crowd that was used to diplomatic subtlety.
Twitter’s removal of Trump a few days after the Jan. 6, 2021, protest and riot at the U.S. Capitol appears to be one of those instances where Twitter executives acted on their own, breaking the platform’s content policies in suppressing the voice of a sitting American president, internal documents indicate.
Twitter suspended Trump’s account on Jan. 8, 2021, after the president made two posts.
“The 75,000,000 great American Patriots who voted for me, AMERICA FIRST, and MAKE AMERICA GREAT AGAIN, will have a GIANT VOICE long into the future. They will not be disrespected or treated unfairly in any way, shape or form!!!” said one of Trump’s tweets.
“To all of those who have asked, I will not be going to the Inauguration on January 20th,” read the other.
Twitter moderators and supervisors agreed that the tweets didn’t violate any rules.
“I think we’d have a hard time saying this is incitement,” wrote one staffer. “It’s pretty clear he’s saying the ‘American Patriots’ are the ones who voted for him and not the terrorists (we can call them that, right?) from Wednesday.”
Higher executives, under pressure from their many anti-Trump employees, wouldn’t accept that conclusion and continued to push for construing Trump’s comments as malicious.
Undermining the Nunes MemoIn January 2018, then-Rep. Devin Nunes (R-Calif.) submitted his memo detailing FBI surveillance abuses in pursuit of the Trump–Russia investigation. The memo was correct on virtually all points of substance, as later confirmed by DOJ Inspector General Michael Horowitz.
The memo was dismissed by the corporate media as a “joke,” but gained significant traction on social media nonetheless. Legacy media and several lawmakers then came out claiming the memo was boosted online by accounts linked to Russian influence operations.
However, Twitter found no evidence of Russian influence behind the #ReleaseTheMemo hashtag.
The claims were all sourced to the Alliance for Securing Democracy (ASD), a group set up in 2017 under the German Marshall Fund, a think tank funded by the American, German, and Swedish governments.
The ASD is closely linked to the U.S. foreign policy and national security establishment. It was headed at the time by Laura Rosenberger, a former Clinton campaign adviser who held various roles at the State Department and the National Security Council. Its advisory council includes former Clinton campaign chairman John Podesta, former CIA head Michael Morell, and former Department of Homeland Security (DHS) head Mike Chertoff.
Twitter officials were at a loss as to how the ASD came to its conclusions.
In fact, the “dashboard” ASD used to make its claims had already been reverse-engineered by Twitter—a fact Roth didn’t want to disclose to the media.
Twitter tried debunking the story behind the scenes without giving out such details, but to no avail. Initially, reporters ran with the story without even reaching out to Twitter, Roth wrote.
The initial letter on the matter from Schiff and Sen. Diane Feinstein (D-Calif.), the top Democrat on the Judiciary Committee at the time, also came out before Twitter was given a chance to respond, internal messages say.
Twitter tried to stop Sen. Richard Blumenthal (D-Conn.) from piling on with his own letter, but again failed.
“Blumenthal isn’t always looking for real and nuanced solutions. He wants to get credit for pushing us further. And he may move on only when the press moves on,” commented Carlos Monje, Twitter’s then-public policy director, in an internal message. Formerly a Department of Transportation official, Monje returned to the department under the Biden administration.
In the end, Twitter never publicly challenged the Russia narrative.
Aiding Pentagon PsyopsIn 2017, a Pentagon official asked Twitter to “whitelist” several accounts the Defense Department was using to spread its message in the Middle East. Twitter obliged, giving the accounts similar privileges it was reserving for verified accounts.
Federal ‘Belly Button’ of InvestigationThe FBI served as a conduit for other government agencies to pass information to Twitter and ask for favors, according to Taibbi.
In one exchange, FBI cyber head Chan explained that the bureau would funnel to Twitter communications from the U.S. intelligence community (USIC), but other election-related communications would come from the DHS’s Cybersecurity and Infrastructure Security Agency (CISA).
“We can give you everything we’re seeing from the FBI and USIC agencies,” Chan said. “CISA will know what’s going on in each state.”
He then asked if Twitter would like to communicate with CISA separately or if it would prefer to “rely on the FBI to be the belly button of the [U.S. government].”
Twitter executives were surprised to learn that the FBI had agents specifically dedicated to searching Twitter and flagging content policy violations.
Since 2017, Twitter has employed at least 15 former FBI agents, further entangling the agency with the platform. The practice is so common that there was an internal discussion group at Twitter for former agents.
Department of Homeland CensorshipThe DHS has managed to shoehorn speech policing into its mandate to protect critical infrastructure. In January 2017, shortly before leaving the White House, President Barack Obama designated elections as critical infrastructure. The DHS’s CISA then made it its job not only to protect elections from hackers, but also from misinformation and disinformation.
In July 2020, CISA partnered with several private research entities to look for, study, and counter election threats. They called themselves the Election Integrity Partnership (EIP) and included the Stanford Internet Observatory, the University of Washington’s Center for an Informed Public, the Atlantic Council’s Digital Forensic Research Lab, and Graphika, a social media analytics firm.
The Atlantic Council acts as a semi-official NATO think tank. It enjoys a tight relationship with the government, particularly the foreign policy and intelligence community. Its roughly 200-strong board of directors includes seven former CIA heads and a plethora of other high-level national security figures.
In practice, the EIP searched social media for anything it deemed a threat to elections, including opinions casting doubt on election results or processes in a “misleading” way. Such content would then be submitted to social media companies for removal or suppression.
EIP leader Alex Stamos said the group was set up to “fill the gap” in countering election disinformation that the government wasn’t authorized to address.
The government isn’t allowed to interfere with the lawful speech of Americans, which is protected under the First Amendment.
The impact, however, was likely much broader. Fist of all, the flagged social media posts were often the key, influential ones. On Twitter, the flagged posts were retweeted, “quote tweeted,” or commented on nearly 6 million times.
Also, EIP made clear that the lists of flagged content often pertained to broader “incidents” or “narratives” and “these lists were typically not comprehensive, but intended to highlight a few examples should the platforms decide to investigate further.”
Information leaked from social media or released through the Twitter Files indicate that the platforms operate both content and keyword blacklists and at least some of them incorporated election-related issues into these blacklists. It’s not known if some such blacklisting was spurred by EIP’s efforts and what was the impact of it.
EIP identified nearly 15 million tweets that, based on keywords and time frame, were related to the thousands of pieces of content it flagged. It’s not clear how many of these 15 million were removed or suppressed.
Furthermore, it appears some of the flagged posts pertained to external content, such as online articles or videos. If social media flagged such external content, all posts containing it would get removed or suppressed and any user trying to later post or link to it would get penalized. For instance, Twitter accounts of the New York Post and even then-White House spokeswoman Kayleigh McEnany were locked for posting the blacklisted link to the Post’s Hunter Biden laptop story.
On Twitter alone, EIP identified over a million tweets and retweets of the URLs it flagged. But again, it’s not clear how many of these tweets and retweets were removed or suppressed.
Generally, the same actors participating in the EIP have also been involved in the suppression of information related to COVID-19 and the 2022 elections.
CISA tried to distance itself from the EIP’s censorship role, saying it never sent the EIP any examples of potential misinformation.
“CISA does not censor speech, period. CISA’s mission is to build resilience to disinformation and foreign malign influence activities that threaten critical infrastructure, including election infrastructure,” a CISA spokesperson told The Epoch Times via email.
“We work in a non-partisan manner with state and local election officials to equip the American public with accurate information about the conduct and security of their elections. Online content platform operators, as always, make their own decisions regarding the content on their platforms.”
A CISA official told The Epoch Times that the agency doesn’t send content removal requests to social media.
This assertion seems to clash with other publicly available information.
A Shift in PrioritiesMike Benz, a former State Department official who handled the cyber portfolio in the Trump administration, traced the government–private censorship apparatus to the foreign policy regime change infrastructure.
The growth of social media in the 2000s was perceived by the establishment as positive because it proved invaluable in accelerating insurgencies. The United States has a long history of supporting local opposition to rogue regimes and dictators, and social media allowed such groups to quickly organize mass protests, as demonstrated by the Arab Spring uprisings, Benz argued.
To that end, free expression online was backed by the U.S. foreign policy establishment.
After the 2016 upsets of Brexit and the election of Trump, however, the establishment soured on free speech. Both events were seen as undermining NATO, and both were blamed on foreign influence on social media—specifically Russia. The U.S. and UK governments in particular saw the need to identify and purge Russian influence operations online and set up a government–private apparatus to do so.
Yet, as the Trump–Russia investigation turned out to be a dud, the establishment crowd had to acknowledge that it was primarily domestic forces driving the populist message, Benz noted.
From that point on, the apparatus set up to root out foreign influence appears to have expanded its focus to target domestic speech instead.
The government, however, cannot target lawful domestic speech openly, since the Constitution prohibits government interference with the political speech of Americans. Despite this obstacle, the government found ways to support the domestic censorship ecosystem indirectly, Benz argued.
One method Benz identified is the issuing of grants to academic institutions to research misinformation and develop methods to counter it.
The National Science Foundation, a federal agency that funds nonmedical research, has given nearly $40 million to 42 U.S. universities to counter “disinformation” or “misinformation” since the start of the Biden administration, Benz discovered.
A $3 million grant went to two members of the EIP, the University of Washington’s Center for an Informed Public, and the Stanford Internet Observatory.
A $300,000 grant went to George Washington University specifically to counter the “populist” messaging of politicians in the United States and several other countries.
Benz became so perturbed by the encroaching censorship that he founded a group called Foundation for Freedom Online dedicated to restoring freedom of expression on the internet.
And there are signs that public exposure of the censorship machinery has had an effect.
The Biden administration was forced last year to put on hold the DHS’s planned Disinformation Governance Board. Its freshly nominated head, Nina Jankowicz, resigned.
“Tell your parents, tell your wife and the kids at the dinner table, tell your friends. If you’re taking rests in between ping pong games, play these YouTube videos for people. All of this starts at the guerilla level.
“This stuff speaks for itself. It will be a cult classic just by virtue of the fact that it exists now. Because once you see this, I’ll guarantee you, you will see this everywhere. This stuff is like a secret decoder ring that will help you slice through the daily news items you see in the censorship space.”
The Twitter files, he said, serve as the “one hole ... that is breaking this Titanic ship of censorship.”