Social media action over a controversial New York Post article on Joe Biden has underlined the difficulties faced by Facebook and Twitter in tackling misinformation, less than three weeks before the US presidential election.
The tabloid published screenshots of e-mails that were allegedly illegally retrieved from a computer containing personal data of Joe Biden's son, Hunter.
The emails revived accusations from Republicans that Biden had helped a Ukrainian gas group, Burisma, evade corruption investigations.
However, the veracity of the New York Post article and how the newspaper obtained these emails came under severe questioning from fact-checkers.
For example, the article states that the FBI was made aware of the information and took possession of the computer, but no federal official has confirmed this.
Meanwhile, a spokesperson for Joe Biden's campaign has pointed to previous Republican-led Senate committees which found that Biden engaged in no wrongdoing related to Ukraine.
The disputed contents and nature of the article presented a conundrum for social media companies, and the following actions from Facebook and Twitter have generated debate.
What did Twitter and Facebook initially do?
On Wednesday, Twitter quickly flagged the article and began banning its users from sharing links to the article in tweets or direct messages.
"We can't complete this request because this link has been identified by Twitter or our partners as being potentially harmful," read an error message.
The New York Post's account remained inactive for several hours, and their original posts sharing the article were suddenly removed.
When the article text was later published on the website of the House Judiciary Committee, Twitter prevented users from sharing the link.
Meanwhile, Facebook's Communications Manager Andy Stone questioned the article's facts and announced that, until the information could be verified, Facebook would reduce its visibility.
The move was particularly surprising, given that Facebook CEO Mark Zuckerberg has repeated several times that social networks should not be "arbiters of truth".
What was the reaction?
The decision to restrict the article's accessibility online generated fiery criticism from US President Donald Trump and members of the Republican Party.
"Terrible that Twitter and Facebook have withdrawn the article about e-mails," Trump tweeted.
At a rally on Thursday, the US president also accused Twitter of blocking the account of his spokesperson, Kayleigh McEnany, for sharing the New York Post article.
Speaking in Iowa, Trump hinted that a US law which largely shields social media companies from being sued over what its users' post, could be repealed.
Trump has previously indicated that he wants social networks to be treated as publishers if they exercise control over what their users post.
For their part, the New York Post also released an editorial that accused social networks of "bias" and "censorship" in favour of Joe Biden's campaign.
"Facebook and Twitter are not media platforms. They're propaganda machines."
The newspaper also argued that hacked emails of Hillary Clinton, the Democratic candidate in 2016, were distributed on the Internet by anonymous sites and WikiLeaks.
What policies did the New York Post article violate?
The decision to restrict the publications of one of the most widely-read daily newspapers in the United States was especially unusual, and Twitter CEO Jack Dorsey soon responded to the criticism.
"Our communication on our actions regarding the New York Post article has not been great," he tweeted on Wednesday evening, while not questioning Twitter's merits.
"And blocking the sharing of the article's web address with zero context as to why: unacceptable," Dorsey wrote.
In a series of tweets, the company then explained that it had blocked the sharing of the New York Post article because it violated two of the platform's rules.
The screenshots of emails included personal data, such as emails and phone numbers, which Twitter does not allow its users to publish.
Moreover, the social network does not allow accounts to share links or images of material that are potentially hacked.
"We don’t want to incentivise hacking by allowing Twitter to be used as distribution for possibly illegally obtained materials," the company said.
Twitter said it would allow users to comment and discuss hacked materials and articles, but they could not include links or extracts of pirated material.
'Disturbing' and 'dangerous' moves
But the initial decision to block and limit the New York Post article has even generated concern among fact-checkers.
Cristina Tardaguila, associate director of the International Fact-Checking Network, said she considered Facebook’s decision to take action without disclosing its methodology "disturbing."
"This is one of the most dangerous moves I've seen in the battle against disinformation," Tardaguila added about Twitter's explanation.
For other observers, the two platforms were misguided in imposing severe restrictions on the New York Post article.
"It gives the story more resonance than it would otherwise have had," tweeted political scientist Ian Bremmer, a professor at Columbia University.
"It is difficult to make a credible case for this particular article to be treated in this way."
Milton Mueller, a professor at Georgia Tech University, says it is impossible for networks to decide whether investigations on Donald Trump's tax returns or corruption allegations against Mr Biden are relevant.
Platforms should "remove the most aggressive forms of content moderation and make it clear that democracy and freedom of expression are complicated issues and that people have to fend for themselves," says Mueller.
Twitter backtracks - so where next?
Facebook has stuck to its original decision, and the New York Post article continues to have reduced visibility until it has been fact-checked.
But following the backlash in response to its actions regarding the article, Twitter confirmed on Thursday that it would be making changes to its policy concerning hacked information.
On Friday, it u-turned on its initial decision to block links to the New York Post piece, allowing users to post links freely.
Twitter's reasoning was that the private information included in the article had become widely available in the media and on other platforms.
Jack Dorsey tweeted on Friday morning saying the “straight blocking of URLs was wrong” and suggested that Twitter instead should have applied tools like labels.
“Our goal is to attempt to add context, and now we have capabilities to do that,” he tweeted.
However, none of the latest tweets sharing the article have included labels - a tool the platform uses to provide additional context and information or clarifications in situations where people could be confused or misled by the content.
When it comes to future pirated material, Twitter "will no longer remove hacked content unless it is directly shared by hackers or those acting in concert with them," said the platform's legal and policy chief, Vijaya Gadde.
"We will label Tweets to provide context instead of blocking links from being shared on Twitter," she added.
Gadde said the company wanted to address concerns that journalists and whistleblowers could be inhibited from participating in public conversation.
"We are trying to act responsibly and quickly to prevent harms, but we’re still learning along the way," tweeted Gadde.
"We will continue to keep you all updated on our progress and more details as we update our policy pages to reflect these changes in the coming days."
In the past, Facebook and Twitter have regularly been accused of censorship by the right, and are criticised on the left for not doing enough to combat misinformation.
On Thursday, Republican senators announced that they plan to summon Jack Dorsey on October 23 to explain Twitter's actions.
The CEO must "explain why Twitter is abusing its power to silence the press and cover allegations of corruption," Republican Senator Ted Cruz said.
Mark Zuckerberg has also been asked to testify "at an undetermined date" before a Senate subcommittee, according to letters addressed to the two social media heads.
Social media companies have sought to develop strong policies for nearly four years after online discord flooded platforms around the 2016 US election and every new step they take is followed by intense scrutiny.