Social media companies bent their own rules in favour of former President Donald Trump and his supporters in the volatile weeks preceding the 2021 attack on the US Capitol, according to an unpublished staff report from the House’s Jan. 6 committee.

The 122-page “Summary of Investigative Findings,” which was obtained and posted by the Washington Post, analyzes testimony from employees and documents received from tech companies including Twitter Inc. and Meta Platforms Inc. It details how the companies failed to adequately address the calls to violence and election-related conspiracy theories circulating online after the 2020 presidential election.

“Both Facebook and Twitter faced significant headwinds in taking aggressive action against problematic content by President Trump and his supporters, partly out of fear that they would be classified as overly partisan,” according to the report.

The committee’s findings contradict complaints from conservative politicians that social media companies are biased against right-wing users. Leaders of the new Republican majority in the House have vowed to investigate allegations of ideological discrimination, using congressional hearings and, if necessary, subpoenas to probe content moderation decisions.

The leak of the document comes as Facebook’s parent company, Meta, decides whether to allow the former president back on its platforms. Twitter reactivated Trump’s account in November after Elon Musk, the company’s new owner and chief executive, polled his followers on whether Trump should be reinstated. Trump hasn’t tweeted since his initial suspension after his supporters attacked the US Capitol two years ago.

The report covers the shortfalls of other platforms, including mainstream companies such as Google’s YouTube and Reddit Inc. as well as those with a smaller user base such as Parler Inc. and Gab. The panel’s investigators interviewed technology industry executives and law enforcement and intelligence officials, along with Trump aides and Republican election officials.

A spokeswoman for Representative Bennie Thompson, the Democratic chairman of the panel, didn’t respond to a request for comment or confirm the authenticity of the document.

The report said that Trump’s Dec. 19 tweet encouraging supporters to come to Washington on Jan. 6 — “Be there, will be wild!” — was a transformative moment across social media, changing the tenor of the conversation online to explicit planning for the event. Plans for violence were telegraphed on both far-right forums and mainstream platforms, which pushed Jan. 6 as a critical day in the Stop the Steal movement, the report said.

The report provides examples of Twitter managers failing to act on employee warnings about heated rhetoric on the platform, including the phrase “locked and loaded.” Employees were told they could escalate any troubling content to superiors, but “no experienced supervisor was on duty the morning of the certification of the Presidential election,” the report said.

The company found its own response to the incitement of violence troubling, according to the report, but “no substantive work was done to improve the situation in 2021,” including in other high-profile elections such as in Brazil last year.

Supporters of former Brazil President Jair Bolsonaro that ransacked government buildings on Jan. 8 this year used tech platforms to cast doubt on his loss in last year’s presidential election.

In the US, Facebook’s Samidh Chakrabarti, then the product lead for the US election, suggested forming a working group to “study the torrent of election delegitimizing content on Facebook.” Guy Rosen, Facebook’s chief information security officer who was running the company’s safety and integrity efforts at the time, told Chakrabarti that even studying the problem “would just create momentum and expectation for action” that he didn’t support.

Meta spokesman Andy Stone declined to comment on the draft report and referred questions about Trump’s reinstatement to the company’s previous public statements, which pledged consequences for future violations if Trump’s accounts are reinstated. Twitter, Parler and Gab didn’t immediately respond to requests for comment.

“We have long established policies that prohibit hate speech, harmful conspiracies and incitement,” Ivy Choi, a YouTube spokesperson, said in an email. “As a direct result of these policies, even before Jan. 6 we terminated thousands of channels, several of which were associated with figures related to the attack, and removed thousands of violative videos, the majority before 100 views.”

Reddit, in a statement, said it found no evidence of coordinated calls for violence on it platform related to Jan. 6.

“Reddit’s site-wide policies prohibit content that promotes hate, or encourages, glorifies, incites, or calls for violence against groups of people or individuals,” the company said in the statement. “We will continue to monitor and enforce our policies across the platform.”

The House panel that investigated the Jan. 6 Capitol insurrection focused in televised hearings and in its final report on the role Trump and his efforts to overturn the presidential election. That 814-page report, which was issued in December, concluded Trump was “the central cause” of the 2021 mob invasion of the Capitol.

The hearings and report included testimony on the importance social media played in spurring supporters to come to Washington and inciting the mob after it penetrated the Capitol to further anger against former Vice President Mike Pence. That included accounts of the impact of Trump’s Dec. 19 tweet, but the panel mostly focused on the conduct of Trump and his associates.