The EU AI ACT (emdash) Article 50: The Last Refuge of Mediocre Journalism?

By David Williams

Is this another example of EU knee-jerk reactionism? It certainly feels like it. The latest phase of the EU’s vast AI regulations arrives wrapped in the familiar language of existential crisis, as though Brussels is heroically defending civilisation itself from an approaching army of rogue chatbots. Under the EU AI Act, which comes into force on 22 August this year, Europe has constructed an enormous framework of compliance rules, transparency mandates, risk categories, oversight mechanisms, and administrative procedures dense enough to make medieval canon law seem positively carefree.

And, predictably, much of the "mediorce" media class has welcomed it with near-religious fervour.

Reading some of the online commentary, you would think artificial intelligence had arrived specifically to destroy journalism, rather than automate a few dreary newsroom functions that were already half-mechanised decades ago. The panic has become unintentionally comic, largely because many of the loudest critics seem never to have actually worked inside a newsroom. They speak about journalism as though it were once produced by candlelight, crafted by solemn intellectuals in tweed jackets, chiselling truth into marble tablets before the silicon barbarians breached the gates.

In reality, journalism has always relied on forms of artificial intelligence. We simply used to call them copy editors.

Every reporter with genuine newsroom experience knows this. Stories have always passed through invisible layers of correction, trimming, fact-checking, rewriting, and occasional emergency surgery performed by exhausted or inebriated sub-editors at two in the morning. The only meaningful difference now is that the machine does not disappear for forty minutes to smoke outside while insisting it is “reworking the lede.”

And frankly, I have had far more trouble with semi-literate copy editors convinced they were future Pulitzer winners than with any AI-assisted fact-checking system. At least the algorithm does not suddenly insert its own ideological manifesto halfway through a story because it once attended a media studies seminar in East London.

The hysteria also flatters journalism with a grandeur it rarely possessed. Much of modern reporting already consists of recycled agency copy, rewritten press releases, Twitter paraphrasing, and journalists quoting other journalists quoting Reddit threads quoting anonymous TikTok accounts. AI did not invent this culture. It merely automated the bureaucracy of it.

There is something deeply Victorian about the panic. One imagines editorialists warning that “the mechanical mind” will destroy civilisation, much as earlier generations trembled before the typewriter, the telephone, or women reading novels unsupervised.

Of course AI can produce nonsense. So can journalists. The archives are full of catastrophically wrong headlines written entirely by humans with degrees, expense accounts, and inflated self-importance. Human error, unlike AI error, usually arrives wrapped in ego.

The real issue is not whether AI belongs in journalism. It always has, in one form or another. The question is whether journalism still possesses enough intellectual seriousness to supervise the tools it uses. A competent editor using AI is infinitely more valuable than an incompetent editor armed only with nostalgia.

Much of the outrage feels less like a defence of journalism and more like a middle-management panic attack. AI is not replacing great reporters. Great reporters are already rare. What it threatens are the vast layers of mediocre institutional gatekeepers who spent decades confusing word counts with talent.

That is the truly unsettling part. AI has exposed something the industry preferred not to admit: a surprising amount of professional journalism was already semi-automated long before the computers arrived.