Chief executive Mark Zuckerberg said Friday that Facebook will remove posts that incite violence or attempt to suppress voting - even from political leaders - and that the company will affix labels on posts that violate hate speech or other policies.
The moves amount to major concessions amid public pressure, employee unrest and a burgeoning advertiser boycott over Facebook's long-standing refusal to more aggressively address hate speech and other platform violations from politicians such as President Donald Trump.
The shifts are at least a partial retreat from the company's traditional deference to speech it deems "newsworthy." That includes Facebook's decision this month to not label or remove a post by Trump that said, "when the looting starts, the shooting starts." Other companies, such as Twitter, which affixed a warning label on a similar post, have been more forceful at responding to what they deemed to be policy violations, including from politicians.
"There are no exceptions for politicians in any of the policies that I'm announcing today," Zuckerberg said in a town hall that was streamed live Friday.
The announcement did little to cool complaints from civil rights leaders, who say they've spent years trying to get Facebook to understand the seriousness of the problems on the platform and had won only modest concessions. They noted that Facebook already supposedly had strict policies against voter suppression and hate speech and that Friday's announcement did little to further address those issues.
"Facebook is feeling pressure, which is good," said Brandi Collins-Dexter, senior campaign director at Color of Change, an activist group long critical of Facebook. "I still think, at the end of the day, they still have a long way to go."
The most consequential change may be Facebook's new willingness to affix warning labels on problematic posts - a step that Zuckerberg long has resisted. The implications could reverberate far beyond the United States at a time when political leaders in many other nations have been exploiting the latitude Facebook has traditionally offered them to lie, misinform and engage in hateful characterizations of other people.
Social media companies are under an especially bright spotlight this year in the lead-up to the 2020 presidential election, facing pressure to control hate speech and misinformation on their sites - something that still haunts them from rampant disinformation that spread online during the 2016 campaign.
Facebook in particular has faced harsh criticism in recent weeks for its decision to leave up posts from the president that many advocates said clearly incited violence.
In the May post, Trump referred to protesters as "THUGS" and wrote, "Any difficulty and we will assume control but, when the looting starts, the shooting starts."
The new policies are not retroactive, meaning Trump's controversial posts over the last several weeks will remain untouched.
But even if they were posted again tomorrow, Facebook's updated policies would not apply to the May post, Facebook spokesman Tom Reynolds said. That post applies to a different policy involving "state use of force," which Facebook is currently reviewing but has not yet announced any changes for, he said.
"A handful of times a year we make a decision to leave up content that would otherwise violate our policies because we consider that the public interest value outweighs the risk of that content," Zuckerberg said in the town hall.
Facebook for years has been wrestling with how to enforce its policies against hate speech, disinformation and other violations when the person posting the content is Trump or some other political leader. As the company massively ramped up its teams for detecting and acting against content that violated policies - hiring tens of thousands of people in the process - it explicitly carved out an exemption for posts or advertisements from politicians, even though they had emerged as a leading source of disinformation and other problematic content in many nations.
The White House and the Trump campaign did not immediately respond to requests for comment.
Zuckerberg has repeatedly expressed his reluctance to have Facebook serve as an "arbiter of truth" and has worked to neutralize claims by conservatives that the company was biased against them and their ideas, even though no systematic evidence of such bias has ever emerged. Zuckerberg for weeks defended his company's decisions, even appearing on Fox News and telling employees the tweet on looting did not constitute a policy violation.
But Friday's action suggested that Facebook's balancing act had grown untenable in light of intense and visible employee backlash, including some high-profile departures, as well as the advertiser boycott.
That it has grown to include in recent days such prominent corporate staples as Verizon, Hershey, Coca-Cola and Unilever is a worrying sign for a company that generates its multibillion-dollar profits and massive stock market valuation by maintaining a robust flow of advertising dollars. Its stock price fell by more than 8 percent Friday, far steeper than the market overall on a down day.
While Facebook's business prospects have bounced back before from major controversies over privacy and disinformation, it has struggled to recover its previously strong reputation among consumers and encountered increasing hostility in Washington from leaders of both major parties.
Facebook said Trump's post on mail-in ballots, which Twitter labeled with a fact-check notice, does not violate Facebook's policies on voter suppression. All posts that mention voting in any way going forward will receive a label to point users to more information about voting, but Trump's posts would not have received a "newsworthiness" warning.
Facebook's Reynolds said that in the future any posts that do violate the voter suppression policy, from the president or otherwise, will be removed. "Our voter suppression policies apply to everyone, including politicians," he said.
The unwillingness to act against such posts remains a major source of tension with civil rights leaders, who for years have lobbied Facebook to take issues of systemic racism more seriously on the platform.
The company announced a policy in November banning posts that suppress voting - even if they come from politicians. Civil rights leaders who worked on that say Trump's recent posts bashing mail-in voting as somehow equivalent to fraud violated that policy and should have been removed because of the potential to confuse voters and potentially dissuade them from participating in a legal means for casting ballots.
Overall, said Vanita Gupta, president of the Leadership Conference on Civil and Human Rights, a Washington-based umbrella group, Zuckerberg hasn't gone nearly far enough in announcing effective policies or enforcing them appropriately.
"They're useful. They're good. But they're super incremental," Gupta said. "They miss the bigger picture of the weaponization of the platform and undermine the prospects of a fair election in November."
Issues related to hateful and violent speech have crystallized in recent weeks amid rising national unrest about police violence. Trump's bid for reelection - following his shrewd and effective use of the platform four years ago - remains a major source of anxiety within the company.
As Facebook has avoided confronting Trump, rival Twitter has started more aggressively labeling posts from him and other politicians in the past few months. It has slapped some sort of warning label on five of Trump's tweets. That also includes one with a doctored video and one that said protesters would be met with force if they tried to set up an Autonomous Zone in the District of Columbia.
Trump lashed out at social media companies over Twitter's labeling, signing an executive order that sought to open the door for a crucial law, Section 230, to be rethought. The law ensures that social media companies are exempted from legal liability for nearly everything their users post on the sites.
In the last week, a growing number of advertisers have pulled their ads temporarily from Facebook as part of the civil rights group-led Stop Hate for Profit campaign, which urges advertisers to put financial pressure on Facebook to implement stricter policies against hate speech.
Facebook said its changes Friday were made after discussions with civil rights groups.
Facebook will prohibit more hate speech in ads, including political ads, such as claims that people from a specific group - of the same race, immigration status or sexual orientation - are a threat. The ads policies will also prohibit language that suggests refugees, immigrants or other groups are inferior in any way.
The company will also point users to official information on voting when they post about the topic and remove posts that try to intimidate or suppress voters.