We’re not ruled by the robot overlords yet

There’s a real early-2000s vibe about AI at the moment.  

That turn-of-the-century moment when people were still clutching onto Yahoo! despite a new technology’s (Google) clear march to dominance, and social media sites were utterly and completely chaotic.  

And now we’re living through it all over again, with a new technology on the ascendancy and a real wild west emerging as the media gets to grips with how it will shape our lives both professionally and personally. 

So, its no surprise that mistakes will be made. It’s just they’ll be realised far quicker than before – particularly in journalism which the internet seems to delight in calling out.  

But I do think there’s merit in learning from the growing number of case studies of where AI in journalism has gone wrong, however agonising it is.  

The bellwether for fake, AI-powered journalism was undoubtedly the infamous Victoria Goldiee case from November 2025. The Guardian and a host of other sites were duped into running “her” content, only belatedly spiking the stories once it became clear that they were entirely generated by ChatGPT.  

But even as far back as 2023, sites were having to pull content due to error-strewn pieces going live. CNET famously had to issue corrections for 53% of the 77 stories written by its AI tool, showing just how fragile the technology was.  

PressGazette unearthed a brazen ring of fake, AI-driven “experts” who were successfully gaming a media request site in the UK, leading to hundreds of articles being removed from publisher websites.  

But perhaps my favourite case study – more for its extraordinary hubris than anything – was the newspapers caught running a summer reading list with books that did not exist. The Chicago Sun-Times even said that it was “inserted into our paper without review from our editorial team.” Wild.  

While these are extreme examples, I’m absolutely not trying to throw any journalists under the bus here. Newsrooms are shrinking and the traditional media is under more pressure than ever. Turning to tools to help ease the stress of that increased workload is natural and understandable.  

But as a result of journalists getting raked over the coals for their mistakes, they’re now on high alert and their tolerance for generic, AI-written pitches from PRs is low.

While AI is a handy tool for everyone, relationship building and targeted pitching is more important than ever in this new frontier. 

Note: PressGazette are running a live tracker of the most notable AI mistakes in journalism. See here.