This week, the Wire published a damning report about the content moderation policies of Meta, the social media behemoth that owns Facebook, WhatsApp and Instagram. Meta, which has misled about how it functions and moderates content before, replied saying the news report was misleading at best and fabricated at worst. The Wire disputed Meta’s claims and published an internal email in a follow-up story. As everybody on the internet started to opine on who had the right of it, things started to get hazy.

So, what exactly happened?

On October 6, 2022, the Wire reported that a number of satirical Instagram stories by “Superhumans of Cringetopia” were taken down on September 19 for violating the platform’s community guidelines. The account does social and political satire in fairly academic language, but its posts are often critical of Hindutva groups and the Narendra Modi government, with some memes comparing Hindutva with Nazism.

The story published by the Wire focussed especially on an Instagram story which mocked a video showing an Ayodhya resident worshipping an idol of Uttar Pradesh Chief Minister Adityanath in a temple dedicated to him.

The reason? The video supposedly violated Instagram’s “sexual activity and nudity” restrictions. Instagram’s guidelines prohibit nudity.

Wait, nudity?

Yes. A Meta spokesperson confirmed to Newslaundry that “nudity” was indeed the reason. But the video showed no nudity or sexual content. Both the idol and the devotee were fully clothed.

Algorithmic takedown or not? 

According to the Wire’s report, this Instagram story, among other posts, were taken down within minutes of being uploaded, raising questions about whether these instances of moderation were the handiwork of algorithms, or if human intervention was involved. If indeed the former, it raises questions about the effectiveness of these AI models. 

In a Twitter chat, one of the people behind the “Superhumans of Cringetopia” account, an Indian national, showed Newslaundry screenshots of seven stories taken down by Instagram between March 7 and September 19 for going against the platform’s guidelines. All notices are marked “closed” except for the Adityanath story which has been “reviewed” twice. The user claimed that all the stories were taken down within two minutes of posting, suggesting it was the work of algorithms.

One of the stories that mocked a post by an Indian user comparing himself to “Dr Joseph Goebbels” after shaving has since been restored, the user said, but its status is still shown as “closed”.

Another satirical post mocking one of Adolf Hitler’s speeches was removed for “encouraging violence or attacks on anyone based on their religious, ethnic or sexual background”.

Satire has been a weak point in automated content moderation since it requires judgement about cultural and political context, something that “Superhumans of Cringetopia” user has pointed out to Meta in the past.

The user, an Indian citizen, declined to get on a call to maintain their anonymity and because they are “already in a big mess”.

Responding to questions by Newslaundry about why the Instagram posts were removed, a Meta spokesperson said, “Content was flagged by our automated systems for review and removed for violating various different policies after being reviewed by our Community Operations team.”

This means the content was first flagged by automated systems and then reviewed by Meta’s in-house human moderators rather than third party moderators social media companies outsource such work to.

To make the matter murkier, the screenshots shared with the Wire indicate that the Adityanath post was reviewed twice. How could this have happened, the Wire asked Andy Stone, communications director of Meta, in an email on October 11?

What does Meta say?

Meta did not respond to the Wire’s email sent on September 28 except to say, via their public relations agency, that they hadn’t received the attachments that the reporter, Jahanvi Sen, sent. Sen told Newslaundry she assumed there was some glitch when Meta forwarded the email received on press@fb.com to the PR agency but sent the attachments again anyway.

That’s the chronology.

But why is all this such a big deal? Aren’t arbitrary takedowns of content common?

Yes, such arbitrary content takedowns are common, as are news stories about them. Sen said when they did not get a reply from Meta about the takedown, they contacted their sources inside the company to find out what had happened. And what they were told basically confirmed what has been previously reported: that India’s governing Bharatiya Janata Party enjoys undue influence on Meta’s content moderation policies likely to the detriment of free speech and fair elections in the country.

Sen’s report hinged on a purported Instagram post-incident report whose log details showed that the Adityanath post was taken down without review because it was flagged by Amit Malviya, the head of the BJP’s IT cell, who is said to enjoy XCheck privileges. XCheck is a controversial Meta programme which allows influential users, including politicians, filmstars, journalists, and sportspersons, much more leeway on the company’s platforms than common users. Their content that violates guidelines usually doesn’t invite immediate action apparently so that the company does not have repeated PR crises on its hands.

Facebook maintains that it is intended to allow content that’s newsworthy and for which authorities should be criticised to remain on the platform but, as the Wall Street Journal reported last year, it has enabled influential people such as Brazilian footballer Neymar to remain on the platform despite violating community guidelines. Neymar had posted nude photos of the woman who accused him of rape but no action was taken against him despite “non-consensual sexual imagery” being banned on the platform.

Sophie Zhang, a Facebook whistleblower, had previously revealed that the social media giant allowed a network of fake accounts to remain on the platform in order to inflate the popularity of a BJP MP.

Stone, Meta’s communications chief, contested the veracity of the Wire story. He said XCheck did not enable people to report posts and that the “underlying documentation appears to be fabricated”.

The Wire responded with another story. It cited a purported email by Stone asking “how the hell” the internal report “got leaked” and telling his team to put Sen and Siddharth Varadarajan, the editor of the Wire, on a “watchlist”.

Not long after, Guy Rosen, the chief information security officer of Meta, tweeted that the two stories by the Wire carried “untrue claims about Meta’s content moderation operations and process” and called them “fabrications”. He also called the screenshot of Stone’s purported email a “fake” saying the cited email address wasn’t current and the “to” field wasn’t in use at Meta either. The post-incident report cited in Sen’s story about Malviya’s XCheck privileges, Rosen further claimed, “appears to be a fabrication”. He also denied the existence of a watchlist of journalists and went on to suggest that the Wire may have been the “victim” of a “hoax”.

Soon, journalists, academics, whistleblowers, even Meta’s critics started raising questions about the Wire’s reportage.

Does Amit Malviya actually have special privileges? How many Indians do?

Who knows.

“We don’t share the list of people on the Cross Check program externally,” the Meta spokesperson told Newslaundry. As to how many Meta users in India have XCheck status, the spokesperson said, “This information is not public.”

According to internal documents leaked to the the Wall Street Journal by whistleblower Frances Haugen, there are about 5.4 million Meta users with XCheck tag globally.

The spokesperson had “no input to share” about how many, if any, XCheck tags have been withdrawn in India, and for what reasons.

Why are there doubts about the Wire’s reporting?

Zhang, a critic of Facebook’s practices, told Newslaundry that in the light of her own experience working at the company as also her conversations with its current employees, “a lot of what the Wire described was improbable and impossible”. She said she believed the “documents are likely fake” and went so far as to suggest that the Wire “should retract the article”. Zhang has previously written an opinion piece for the Wire.

Zhang, however, didn’t discount the possibility that this could have been a psyop to discredit the Wire, which has a record of reporting critically on the governing party.

Alex Stamos, a former chief security officer of Facebook who now heads Stanford University’s Internet Observatory, claimed “the Wire just destroyed their credibility”, and that “this could have been an extremely successful op against opposition journalism”, presumably by the BJP.

He also claimed that, in his experience, 90 percent of Stone’s email replies are to take the conversation offline.

Shoshana Wodinsky, an American journalist at MarketWatch who has written extensively and critically about Facebook and its privacy practices, also highlighted several problems with the Indian news website’s stories.

Another point that critics, including Stamos and BJP critic Aakar Patel, have pointed out is that the language used in the email does not “sound” like that of a native American speaker.

For internal emails, Meta uses Outlook 365. For internal chats, it uses its own Slack-like platform called Workplace.

For external emails, press@fb.com is the address the company’s communications team still directs journalists to. It’s conceivable, as Rosen notes, that Stone’s email address cited by the Wire isn’t current, but it does exist. Newslaundry sent an email to that address and it didn’t bounce back. This means that the email address is still active, if not in use.

Newslaundry asked Sen, the Wire journalist, if at any point they thought Stone’s purported emails were too good to be true. She said, “No, because we are confident in our sources.”

What verification process did the Wire follow?

It was only after the Wire had published the first story about the content takedown on October 6 that one Meta source told them what had happened. The Wire asked the source if there was any internal report about the takedown and the source provided the post-incident report, which the website published on October 10. Sen declined to share their mode of communication to protect the source.

The Wire subsequently had at least two other sources in Meta corroborate the report.

On October 9, the day before the story about Malviya being a XCheck user was to run, Sen followed up on her email sent on September 28 to press@fb.com and cc’ed Meta’s PR agency Weber Shandwick and asked about the reasons for taking down the posts by "Superhumans of Cringetopia". This email didn’t mention the post-incident report, XCheck programme, or Malviya’s involvement.

Asked if mentioning the report they had received from the first source could have helped cover their bases, Sen said they chose not to for two reasons. One, to protect the source. Two, given Meta hadn’t responded to their requests for information, they were unsure “if they would take any other information we sent them in good faith”.

At around 9 pm on October 10, after the Malviya story had gone online, Sen received three calls from an unknown number which she didn’t take. “I don’t usually take calls from unknown numbers after work hours,” she said. Then, she got a series of messages on WhatsApp from a Meta communications official. Sen immediately called him back.

In Sen’s telling, the Meta official asked to speak off the record because he wasn’t a spokesperson. He said the Malviya story was misleading, did not report on XCheck correctly, and that XCheck had nothing to do with the takedown of the posts. When Sen asked if he thought that the post-incident report cited in her story was not genuine, he claimed that he was unable to see the document on the website.

The Meta official promised to call again after collecting more information. When Sen told him she needed an on-the-record response from the company, he hung up and never called back.

Sen got the call about half an hour after Rajiv Aggarwal, head of public policy for Meta in India, purportedly emailed Stone saying he had assigned “a comms member from India to talk to the journalist”.

The next morning, Sen followed up with the Meta official on WhatsApp. By then Stone was replying to tweets about Sen’s article and denying the veracity of its claims.

After Stone’s statements, the Wire contacted another source at Meta. At least one employee at the Wire had previously met this source in person and verified their identity. This new source gave the Wire the email purportedly sent by Stone to his team and the publication had two other people confirm its authenticity.

The Wire emailed Stone and the Meta India communications executive asking why they claimed the post-incident report “appears to be fabricated”, why the Adityanath post was reviewed twice if the moderation was only automated, whether Stone had called for an internal review of Meta’s communications with the Wire, and if Sen and Varadarajan had been placed on a watchlist. In neither of these emails was press@fb.com cc’ed.

What is the scene with the watchlist?

Stone’s purported email cited in the Wire’s October 11 article mentions placing Sen and Varadarajan on a “watchlist”. Rosen denied the existence of such a watchlist as did Meta in its press statement on October 13.

TechCrunch, in its story on the saga, reported that Facebook, like other companies, “does maintain dossiers on journalists” and that Manish Singh, one of the writers of this particular story, received one such dossier. Singh told Newslaundry he accidentally received the document by email in 2016, ahead of the launch of WhatsApp voice calling in India. He no longer has a copy but recalled that it had 27-30 pages and each page had a journalist’s name, details of the kind of stories they did, and how Facebook could reach out to them in ways that served Facebook’s communication goals better.

Singh recalled two names on the document, Nandagopal Rajan, new media editor at the Indian Express, and Thomas K Thomas, Mumbai bureau chief of the Hindu Business Line. Singh, however, pointed out it’s a common practice for companies to keep such lists.

So, did The Wire get duped? Were their practices up to the par?

Sen and Varadarajan maintain that they did not get duped and did their due diligence. And going by the process that Sen shared with Newslaundry, they do appear to have done so.

The story, though, is still unfolding.