Fifteen PR professionals, all pitching journalists, all working for the same London agency. None of them real.
That is the story Maggie Harrison Dupré uncovered for Futurism at the end of July, and it spent the weekend doing the rounds of every PR and media WhatsApp group in Europe (before hitting LinkedIn). It deserves the attention, though not for the reasons a cautionary tale usually earns it. What makes this one worth sitting with is how ordinary the mechanics were, and how much damage they did to people who had no idea it was happening on their behalf.
So, what happened?
The reporting concerns the a UK firm representing a roster of tech clients. Dupré began this journey investigating a single pitch from a PR pro that didn’t seem to be real. She noticed that the headshot was machine-generated, and then found that the image still carried the watermark of the site that had produced it (to the lol of many). Pulling that thread, and speaking to colleagues in the world of journalism, she discovered at least fifteen distinct fabricated personas, each with invented names and email addresses, several with AI-generated faces, operating out of a scatter of plausible-sounding agency domains that all redirected to the same real company.
The pitches were not a brief experiment either. Dupré found messages going back to September 2025, close to a year of outreach conducted under names that belonged to nobody.
Tech.eu‘s Cate Lawrence, also writing about the fallout, spoke to the European startup ecosystem and made a point that not all startups may consider. For a journalist, trust and relationships are the entire job. She had received legitimate pitches from the same agency over the years, ones that led to real interviews and real coverage, which is precisely what makes this crazy fabrication so hard to explain. Startups no doubt will experiment, but hopefully not before learning this lesson.
Companies should take a closer look at their agencies
According to Dupré’s reporting, the clients whose stories were being pitched did not know about this. One founder said the practice ran against everything they stood for. Another said they would have objected immediately had they been asked.
It’s getting ethical. Think about what that means in practice if a startup hires an agency to represent it to the press, but goes out under invented identities. And when the story breaks, the startup’s name appears in an article about deception in PR. Not a great look! Companies hiring PR agencies might want to make sure they are aligned on this.
The craziest part is that the deception actually worked. Dupré traced one pitch sent by a fake persona to coverage that later appeared in a national outlet. Imagine this, at the kind of scale AI allows.
Not an isolated failure
It would be comfortable to file this as one agency behaving badly. The reporting suggests otherwise. Speaking to Tech.eu, Gradient Labs chief executive Dimitri Masin, himself a client of the agency, estimated that around seventy per cent of the unsolicited email he now receives is synthetic outreach from fabricated identities, and described it like this: “AI-generated outreach from fake identities is becoming an epidemic”.
We can confirm that ourselves: we get multiple emails daily offering lead-sourcing services, offers to acquire our company, and more, with emails that are clearly automatically written and sent. Time for email filters to level up?
Trust is not a nice-to-have, it is the product
Here’s the most important part. PR runs on trust and on human contact. That applies to the veracity and authenticity of what gets shared, and equally to the manner in which it is shared. The relationship between a PR pro and a journalist is close to sacred, and we do not use that word lightly. Why taint it by sending emails through a fake persona?
And journalism actually runs on exactly the same currency. Readers rely on journalists. Journalists and editors rely on each other. Journalists rely on sources (and trusted PR pros). And the whole arrangement holds together because there is a person at each link who can be held to what they said. Remove the person and there is nobody to hold responsible for a false claim, a misleading figure, or a pitch that should never have been sent. Fake personas and AI slop are a red flag precisely because they remove accountability at the first link.
People read journalism to form opinions, to make decisions, and to stay informed with material they can rely on. For that they trust human journalists most. And journalists, in turn, trust human counterparts.
AI is useful in so many areas. But not as the source of truths in which we put our trust.
So where does AI belong?
The perfect topic for our next blog post, right? But let’s touch on this for a second.
Nothing here is an argument against using the tools. They are useful for research, for interrogating a dataset, perhaps pressure-testing an angle before it goes anywhere near a journalist. For the unglamorous work of admin, research and more. We use AI a lot. Just… in specific areas, and with a clear red line. Which, if you cross, may cost you the trust in your agency and clients.
