The UK’s newly-designated regulator of ‘online harms’, Ofcom, has been setting out, and will for the next year or more, how it will require apps, and online services such as search engines to protect particularly children from many kinds of content – such as material that promotes hatred, bullying, suicide. Those platforms must have ‘content moderation systems and processes’, so as to act on harmful content (such as ‘fraudulent advertising’). Mark Rowe, as part of a series about security-related jobs, looks at content moderation, and how its work throws up work for security departments at those online firms.
The sheer amount of stuff online! Music (does its video soundtracks promote misogyny, or hatred of minorities, or one side or another in the Indian or any other elections?!). Podcasts, blogs, Facebook comments. More stuff than humans can possibly check. Pornography, and promotion of everything wicked known to man by all sides in all conflicts globally. As for saying that artificial intelligence or some piece of tech can manage what other pieces of tech have created, if tech were to moderate what’s posted, the decisions may further enrage people. Hence, as in AI discourse generally, the need for a ‘human in the loop’. A security exec put it to me: “No one stays long,” in content moderation work, “two years max.”
One turns almost as if to another era to Lord John Stevens’ memoir of his years in the police, including as Commissioner of the Metropolitan Police, Not For The Faint-Hearted: My Life Fighting Crime, published in 2005. He recalls that as chief he would get phone calls from people he knew, including journalists, asking for a favour – would he do something to get their burglary better, or more quickly, investigated? We can add as an aside that such demands on the country’s most senior cop need not happen any more, not after last summer the then Home Secretary Suella Braverman got agreement from police chiefs to follow up all ‘reasonable lines of inquiry’.
In the new, online days, the same happens; at least one security exec at a platform has got called up by those he knows, asking for something to be taken down, or at least looked at by content moderators; and the exec has replied that the caller knows perfectly well there’s too many cases to be able to prioritise one.
But staying with journalism. Is a platform like Facebook performing the functions of a newspaper or television station? The platforms deny it because then they would fall under regulation as applies to the news media. Pre-internet, if a newspaper printed something that some reader knew was unfair and inaccurate, or simply didn’t like even though it was fair and accurate. Even if their gripe was silly – the newspaper had not been admiring enough of their favourite pop group, they might go to the physical office and complain. The newspaper office reception would field the complainants and say something to fob them off or wait for them to go away. At least the newspaper, whether daily or weekly, was next day or week’s fish and chip wrapping. Online, if one side in an African ethnic dispute posts hateful things about you and your family, or some business rival traduces another in Timbuktu, it’s there forever. Hence security issues when people with grievances seek to speak to someone in person – and oddly, those complainants may have printed out the content, and have made a file of it.
The security issue here is about the platforms’ corporate offices. They don’t advertise themselves as much as they used to, or as you might think they would, given their size, profitability and reach. A building might have a logo or name so that deliveries and other visitors know where to come; otherwise they’re as discreet as an Oxbridge college. A security exec said that they’d rather their employer was even more discreet on the outside of the building; although, as they went on to acknowledge, you only have to go onto an internet search engine (ironically) to look up an address. Hence people know where to turn up, to protest or merely bend someone, anyone’s ear for seemingly or literally hours about their case. Also relevant here may be a comment in the feature in the May edition of Professional Security Magazine, about corporate security; that a corporate is only two clicks away from a scandal. Whatever you are protesting against, destruction of the Amazonian rain forest, nuclear power, HS2, you only have to go into a search engine, trawl podcast sites, and you will find some corporate sponsoring something, an exec saying something, or the business doing some deal, that makes the corporate complicit in whatever it is you are opposed to. And that angers you enough to find out where that corporate has an office, and protest outside it, and maybe occupy it or break a window and (more irony) post videos of the deed on platforms. More content to moderate!?
