Who’s Sending AI Scam Story Pitches to Newsrooms?

On Jan. 21, three different Tyee editors received three different pitches within three minutes of each other.

The pitches were eerily near identical. At about 140 words, each opened and closed identically and painted a picture of a B.C. story that, broadly speaking, could fit well in The Tyee’s pages.

While they offered analysis, the pitches had no specific details.

Each email was also signed by a generic, rhythmically similar name, a couple of them appearing inconsistently across the email, either misspelled or in reverse order.

The Tyee responded to two of the pitches requesting a portfolio. In response, both sent a link to the exact same portfolio website — with a different byline entirely named on the pieces.

After we followed up about these discrepancies, the emailer said he sometimes used a pen name.

My guess — bolstered by running the text through free artificial-intelligence detectors — was that the pitches and the pieces on the portfolio website were AI-generated. In other words, fabricated and not based on research or interviews with sources.

The Tyee is not the only news organization to have received eyebrow-raising pitches recently. The National Observer, the Local and the Grind have, too. Editors are turning to each other for advice about how to navigate this new reality. One even seemed to have a heart-to-heart with his AI fraudster.

Newsrooms such as the Chicago Sun-Times, the Philadelphia Inquirer, Business Insider and Wired have made headlines for publishing — then retracting — AI-generated content.

It’s enough to give pause to assigning editors and particularly troubling for those whose mandates include giving new and underrepresented writers a chance to prove their abilities.

In the case of the flurry of pitches The Tyee received, we were all but certain the emailer wasn’t looking to do serious journalism. Still, I asked if we could hop on a video call to discuss the pitch, and he said yes.

Looking for answers

I reminded myself to keep a curious tone, not an accusatory one, as executive editor of the Local Nicholas Hune-Brown did in his November account of his investigation into a possible freelance scammer.

After calling for pitches online, Hune-Brown had accepted one from a “Victoria Goldiee.” Her clippings “painted the picture of an ambitious young freelancer on the rise,” he wrote. But there was no trace of the Canadian bylines she claimed.

Then Hune-Brown decided to dig; the expert source Goldiee had cited in her pitch had no clue who Goldiee was. When Hune-Brown confronted her, Goldiee claimed it was because she’d had her assistant do the interview. She also claimed her Canadian bylines weren’t online because the editors she worked with no longer worked at those newsrooms.

After about five minutes, Goldiee, who had agreed to a video call but at the last minute said she’d be available only via audio, hung up on Hune-Brown — and wasn’t heard from again.

In our case, I expected our emailer not to show up to our scheduled video call. But, four minutes late, he did.

The young man appeared to be in a concrete basement, with a backdrop of computer monitors, adapters, large binders and other miscellaneous tech items.

After a quick greeting the audio on the call went fuzzy, and after a minute and a half he finally hung up, citing technical difficulties.

Our second call was more fruitful. The young man said he lived in Nigeria, did not have plans to speak to sources and was a part-time journalist and a part-time web designer.

When I asked him about the second email that linked to the same portfolio website, he said it wasn’t from him. Maybe it was a mistake, he said.

He told me he didn’t use AI to write the pitches or in his journalism. “AI cannot be trusted fully,” he said. Soon, the screen went dark.

When Hune-Brown at the Local was piecing through his experience with Victoria Goldiee, he initially guessed her name was just one out of a larger content farm.

In the end, he shifted his assumption. “I think it’s a real individual,” he wrote.

You might have one lingering question. Why? Why send faked pitches?

Mathieu Lavigne, analytics lead of the Media Ecosystem Observatory, said that since they’re not the ones receiving the emails, they can’t “comment on the motivations or investigate who is submitting many AI-written pitches. It's therefore hard to know whether there is any co-ordination, a political agenda, people trying to build a fake portfolio, etc.”

Hune-Brown guessed it was the oldest reason in the book: quick cash.

If you’re scamming North American journalism outlets from an area of the globe with a much lower cost of living, he wrote, it could result in “a pretty decent payday — especially if you just have to enter it into ChatGPT.”

Avoiding the scam

Journalism is focused, ideally, on disseminating verified information in the public interest.

At a time when reader trust in journalism has been eroded by misinformation, disinformation and social media echo chambers, keeping AI out of journalism is particularly important.

While many outlets have AI policies they share with potential freelancers, when it comes to scams and writers bending the rules there are no definitive tools editors can use to confirm if generative AI was used to research or write a piece. Often, they need to rely on their own judgment.

“Current detection tools are not extremely reliable,” Lavigne told The Tyee. “It can be difficult for newsrooms to draw a clear line between legitimate content and content generated en masse by AI.”

When The Tyee asked for tips on how editors can spot an AI-generated pitch or draft, Hune-Brown said that if a writer is based in a different country from the story they are pitching, that’s a red flag.

“I could tell they did not know what they were talking about, because they were not from here,” he said.

Another red flag is when the writing is “extremely generic.”

“The Victoria [pitch] was unique,” he added, “because it was so full of lies that it made it a little bit more convincing.”

“For my whole career, I've operated under the assumption — which was maybe naive — that the ideas and prose in a pitch are fundamentally attached to the person who sent it. Post-ChatGPT, there's no sense that any one sentence is created by a human.”

Hune-Brown isn’t convinced that “there’s something innately human you can see in the soul of a piece of writing.”

He pointed to a viral New York Times quiz asking readers to choose the better writing sample, where one was written by a human and the other by AI.

The quiz was “very annoying,” he said. “We've reached a point where I'm not counting on my, like, Spidey senses to be able to tell if something’s AI or not.”

In the case of the three pitches sent to The Tyee, delving into the portfolio website offered us more clues. Most of the “published works” on the sites linked to Google Docs. Of those, four were set as private and therefore unreadable.

Only three out of 10 works linked to pieces published on media sites. The writer’s bio differed on each website: one described him as a culture journalist exploring identity, another as a humanist writer exploring science.

As for contact info, the website offered “0000000000” as a phone number, and the email link went nowhere.

Still, some scammers are less sloppy than others.

The impact on freelancers

Once upon a time, editors identified gaps in a freelancer’s credibility through an organic writer-editor relationship. But the era of digital media has made that trickier.

For Jimmy Thomson, editor-in-chief of the National Observer, the issue of scam pitches became a speed bump after posting a story call-out on Bluesky.

He received an unusually high volume of pitches in response. Some were even compelling.

“It was like someone really knew our publication,” Thomson told The Tyee. He recalled one that read: “I’d love to pitch a feature for the National Observer examining how the Airbnb and short-term rental boom is straining infrastructure and ecosystems in rural Indigenous territories across Canada, from Tofino to Haida Gwaii to Prince Edward County.”

Given the details of the pitch, Thomson assumed that the emailer was based in Canada. He assigned the pitch.

But the first draft, Thomson told The Tyee, contained no characters or news hook and used unnamed sources.

Then Thomson ran the draft through five AI detectors, which identified 80 per cent of the draft to likely be AI-generated. He killed the story.

“It went a lot further than I would like to admit,” he told The Tyee. “It's possible we could have published this story. In the end, the story wasn’t published, but it was close.”

When he received the pitch, Thomson said he hadn’t wanted to count the freelancer out just because he hadn’t heard of them.

Upon further inspection, it turned out the email was signed off by the same name that a Washington Post investigation connected to a series of AI-generated articles published, then retracted by Business Insider and a number of other publications.

“It's sad that I'm now far more skeptical of any writer I haven't already encountered in some form or doesn’t come with a recommendation from a known colleague,” he added.

According to a Reuters Institute study published in February, it’s not statistically clear yet how generative AI is changing the freelance game. Some of the 45 freelancers and editors who responded to reporter Marina Adami’s call-out attributed a decline in work to the use of AI, some said they’re getting more work, and some said there has been no change.

“Rising skepticism from editors could create additional obstacles for young journalists or those just getting into the business,” notes Adami, “as editors look beyond a pitch to a writer’s previous bylines or contacts, for example, before assigning a job.”

Neither Thomson nor Hune-Brown has called for pitches since their close calls, instead leaning on writers they know.

It’s a “pretty tragic” reality, Hune-Brown said, adding that writers who didn’t attend prestigious journalism schools used to be able to open doors by packaging a strong pitch for editors they didn’t know.

Reaching for creative solutions

Thomson said he’d like to see some kind of portal, perhaps administered by the Canadian Association of Journalists, where you can verify yourself as a journalist and submit pitches to editors.

“That way you know it’s coming from a real person who does real work,” he said.

This is the biggest issue, short-term, said Hune-Brown: figuring out how to bring new writers onto a publication’s roster. Part of the Local’s mandate is “reaching new writers from underrepresented communities in Canadian journalism,” he added.

These days, the Local has stronger editing and fact-checking processes, he said. “We're getting annotated drafts from all our freelancers.”

Hune-Brown is also returning to the analogue era of editor-freelancer relationships. In other words, getting more coffees with young writers to get to know someone before the pitch process even begins.

For writers looking to connect with editors, Hune-Brown suggested not just reaching out with a pitch but simply introducing yourself.

Although, he added, Goldiee did that months before her first pitch, too.

If you’re a newsroom editor and have received pitches like the ones discussed in this story, get in touch with Sarah Krichel by email.

Read more: Media