Earlier this week, I was asked to list the three biggest threats to the media. Aside from the general sales decline of newspapers, I said, the threat of bot censorship – and the lack of accountability from the firms who apply it. We at The Spectator have just come across a classic example of this, when Facebook refused to publish this week’s cover satirising Joe Biden when we submitted it as an advert. The cover asked if Biden would serve for six more years, but the illustration had him holding up five fingers. A nice joke, but hardly a cruel one. So we appealed. An email came back saying:

You asked for another review of your rejected ads. After another review, it’s been determined that they still don’t comply with our Advertising Policies.

So that was it. Our Biden cover was rejected. No further recourse. Computer says no.

What happened next makes an interesting case study of the power – and accountability – of the social media giants that hold such sway over British public debate. Facebook is now the no.1 source of UK news after the broadcasters, its bots deciding which news posts are promoted and which ones are concealed. Whoever programmes the bots wields more power than any of the great media barons: Hearst, Beaverbrook or Murdoch. I’m not saying that Facebook is politically biased, just that bots are getting more energetic which makes life harder for satire and against-the-grain arguments (from left or right). This is something I now encounter daily: a Kafkaesque process of rejection, lack of explanation and algorithm editing which has far more influence over what we read than is generally acknowledged.

Facebook, like most social media giants, does not feel the need to respond to people asking why their content has been targeted. There is no hotline to call, no account manager to complain to. We always use Facebook ads to promote our cover: over the years we have shown Boris Johnson crashing into the ground with his head splat on the pavement. We have shown Liz Truss and Kwasi Kwarteng in a motorcycle crash, their bodies in mangled metal. We’ve mocked Theresa May as a ‘Maybot’ putting her in all kinds of positions of torture for weeks. And Donald J. Trump? We’ve had him as a gun-toting psychopath and a pitchfork-wielding loon, goose-stepping in synchronisation with Marine le Pen. All published by Facebook without a problem.

But a gentle teasing of Joe Biden and all of a sudden we don’t comply with its ‘advertising policies’. And even more spookily, our ‘registration status’ – whatever that is – has now downgraded from green to amber. Why? We were not notified. It just happened. ‘The Spectator could not be registered. You can make changes and submit again in 60 days,’ says the bot. End of.

When I tweeted out our plight, it caught the attention of newspapers. I received a call from the Sun telling me that Facebook had said our ‘advert’ had been rejected as it was not posted by the right person. This was odd. The only reason we were given was that the problem lay with the advert itself. And while there is an ‘authorised person’ process (our Facebook ads are handled by an agency) magazines are supposed to be exempt from it.

Facebook’s advice says this (my emphasis):

Any advertiser running ads about social issues, elections or politics who is located in or targeting people in designated countries must complete the authorisation process required by Meta, except for news publishers identified by Meta.

The Spectator is, of course, a news publisher – with a blue-tick, no less. Our Facebook page is below:

So this seemed odd. Facebook was changing its story. As newspaper interest in our Biden cover story grew, we had a response from Alex Belardinelli, Facebook’s public affairs guru, a lovely guy whom I knew when he worked in Westminster. ‘Thanks for flagging this,’ he said on Twitter. ‘As you know, anyone who wants to run an ad that’s about politics or elections has to be authorised – per our process here. If The Spectator resubmits it from an authorised page admin the ad will be approved.’

But we had been told the Biden problem was the advert, not the authorisation process. Why was The Spectator being hauled up under a technical FB rule applying to political adverts? As a news organisation, we thought we were exempt from such rules. We have applied for all the various social media statuses: blue-tick verification, business verification, media verification and a waiver disclaimer for running political ads. And none of our (many) Facebook adverts has been a problem – until we mocked Sleepy Joe.

Facebook was now telling us to try to post the advert again. So we did. And guess what? We have been rejected again. We’re appealing again.

Facebook’s processes are opaque, and are designed to be. Its censorship bots work on criteria that no one is ever allowed to know. When Facebook gets busted, it cites a technical flaw. Whoops, sorry we deleted the Socialist Worker’s Facebook page. It’s back online now, no conspiracy! Whoops, sorry about that Joe Biden ad, if you ask an official administrator to publish then we’ll approve next time. (Except they didn’t.)

But what I really wanted to talk to Alex about was why Facebook has slapped a ‘false information’ label on an article The Spectator published by Carl Heneghan, an Oxford professor, about the evidence behind face masks. What information was false? But my emails have gone unanswered. Facebook hides behind independent fact-checkers that they use. But it was Facebook, not a third party, that used the (potentially libellous) words ‘false information’ on Professor Heneghan’s work. So surely Facebook has an obligation – as any journalist would be under – to justify such an incendiary verdict?

The bottom line is that Facebook is a California-based company which censors British content without being answerable to any UK authority. It’s also a company that has created an amazing audience of billions of people, technology that provides about 8 per cent of The Spectator’s traffic and probably our new subscribers too. We need it. We are a stronger magazine, reaching millions more people, because of the audience that Facebook has developed.

This is the dilemma. The ‘public space’ now, is digital: the private property of a handful of tech giants that created it. They are beholden to no one. So they can strike down a sitting president because they feel like it, delete a video with David Davis attacking lockdown rules in parliament etc. There is no regulation of this and no transparency, no obligation to notify publishers whose articles are targeted, leaving us to guess how we offended the bots.

The moral of this story is not that Facebook is censoring anti-Biden material. This shows how all publishers are now fighting blind in a world of bot-censors whose remit we can only guess at. These firms live in fear of publishing fake news, or misinformation – so their bots let rip, take down a lot of accurate and fair comment, and they don’t much care who gets gunned down in the crossfire. They fear that if their rules are clear, then genuine bad faith actors (conspiracy theorists, the Kremlin etc) will find a way around the rules. So they don’t want clear rules. The Kafkaesque process is intentional.

The highest-risk content, now, is satire. It’s against-the-grain opinions, from political writers or Oxford scientists. It’s challenger publishers like Novara Media on the left and TalkRadio on the right. And that’s why these bots represent such a challenge to freedom of expression. The lack of accountability is a real problem and one that will be made a lot worse by what’s coming in the Online Safety Bill.

Meanwhile, if anyone from Facebook is reading this and would like space on The Spectator website to explain why Professor Heneghan’s article constitutes ‘false information’ then it’s an open invitation. And as for the Biden cover: we’re sure we’ll get there eventually.