Facebook has said it will start to label potentially harmful posts that it leaves up because of their news value.
The more hands-on approach comes as the social media firm is under pressure to improve how it moderates the content on its platform, including posts by US President Donald Trump.
More than 90 advertisers have joined a boycott of the site.
Consumer goods giant Unilever on Friday added its name to the list, citing a “polarized election period” in the US.
The maker of Dove soap and Ben & Jerry’s ice cream said it would halt Twitter, Facebook and Instagram advertising in the US “at least” through 2020.
“Continuing to advertise on these platforms at this time would not add value to people and society,” it said. “We will revisit our current position if necessary.”
In a speech on Friday, Facebook boss Mark Zuckerberg defended the firm’s record of taking down hate speech.
He pointed to a European Commission report this month that found Facebook removed 86% of hate speech last year, up from 82.6%.
What did Mark Zuckerberg say?
But he said the firm was tightening its policies to “address the reality of the challenges our country is facing and how they’re showing up across our community”.
He said the firm would ban ads that describe different groups, based on descriptors such as race or immigration status, as a threat. It will also remove content – even from a politician – if it determines that it incites violence or suppresses voting.
Mr Zuckerberg also said the firm will attach a label to “problematic” content that falls outside of those categories.
“A handful of times a year, we leave up content that would otherwise violate our policies if the public interest value outweighs the risk of harm,” he said. “Often, seeing speech from politicians is in the public interest, and in the same way that news outlets will report what a politician says, we think people should generally be able to see it for themselves on our platforms.
“We will soon start labelling some of the content we leave up because it is deemed newsworthy, so people can know when this is the case,” he said.
Twitter has already taken some similar steps, including banning advertisements from politicians and adding labels and warnings to some kinds of content,including tweets by Mr Trump.
“We have developed policies and platform capabilities designed to protect and serve the public conversation, and as always, are committed to amplifying voices from under-represented communities and marginalized groups,” said Twitter executive Sarah Personette.
Shares of Facebook and Twitter both fell more than 7% on Friday.
Some boycott organisers said Mr Zuckerberg’s promises did not go far enough.
“What we’ve seen in today’s address from Mark Zuckerberg is a failure to wrestle with the harms [Facebook] has caused on our democracy & civil rights” said Color of Change president Rashad Robinson said.
“If this is the response he’s giving to major advertisers withdrawing millions of dollars from the company, we can’t trust his leadership,” he wrote on Twitter.
Why are companies boycotting Facebook?
The “Stop Hate for Profit” campaign was started by US civil rights groups after the death of George Floyd in May while in police custody. It has focused on Facebook, which also owns Instagram; and WhatsApp and last year attracted advertising revenue of almost $70bn (£56.7bn).
The organisers, which include Color of Change and the National Association for the Advancement of Colored People, have said Facebook allows “racist, violent and verifiably false content to run rampant on its platform”.
More than 90 companies including Verizon and Patagonia have joined the campaign, according to a list by ad activism group Sleeping Giants, one of the organisers.
Nicole Perrin, analyst at eMarketer, said it will be difficult to determine the financial impact of the boycott on Facebook, given the significant changes in advertising amid the pandemic.
But she said Unilever’s announcement was significant, noting that the firm was dropping the ads for longer than called for, and on more platforms.
“That suggests a deeper problem with user-generated content platforms, as divisiveness is to be expected on any such platform that allows political expression,” she said.