Industry Impact3 min read

A Real Writer's Track Record Can Now Be Borrowed by a Fake

By , Senior AI ConsultantPublished

Someone using AI got a paid article into Science under a real journalist's name, which shows that checking a stranger's past work no longer proves who they are.

Three freelance science journalists, Linda Geddes, Caroline Williams and Sara Novak, have each found a stranger pitching stories in their names. Geddes also found an article in the New York Post under her byline that she never wrote, with quotes lifted from her earlier features. The email addresses behind the three impostors appear to share the same hidden recovery address, which suggests one person or one group. Nobody knows who.

An invented writer leaves a trail, a borrowed one does not

Every step of this scam is nearly free. The real writer's published articles are public, so the pitch can name them and sound like a colleague. An AI writes a fluent pitch and a fluent first draft in minutes. The editor's usual question, whether this person has done real work before, gets a true answer, because someone has. The only step that costs anything is getting paid, and in Williams's case that worked: Science published the piece in June and removed it three months later, once editors who had all received odd pitches compared notes.

That is the difference from the Blanchard case. A fake with no history can be caught by looking for a history. A fake with a stolen history has to be caught by something outside the writing, because the writing is now good enough, and by something outside the name, because the name is real.

The new check lands on newcomers

Editors have settled on a call. Science requires a video call with any new writer, and an editor at Nature Health is wary of commissioning anyone he has not worked with. At a different magazine, Undark, a pitch with no publishing record behind it collapsed on a phone call, because the writer could not talk about the material in her own pitch.

A call catches the lazy impostor, but it is a stopgap. In February, hackers linked to North Korea ran a fake Zoom call that showed a deepfaked video of a trusted executive, built to get the target to run a harmful command. A call is better than an email, but it is still a channel the other side can fake.

The people who pay most are the honest beginners. A writer with ten published pieces can be checked in a minute. A writer with none is now a possible fraud until a call says otherwise, and every editor who has been fooled once will ask for the call every time.

The same swap is coming for hiring and buying

Nothing here is specific to magazines. A recruiter already sees CVs that are polished and tailored, and criminals have used stolen personal details to get remote jobs under other people's names. A small design agency that takes on a freelancer from an online application could meet the same case: a portfolio of real work, a plausible cover letter, a real person's name, and nobody behind it that the agency has ever spoken to. Even a name on a scientific paper can be borrowed, as Eric Topol found when a fake journal listed him as an author in April.

A call of fifteen minutes with each of 20 new people a month costs a manager five hours. That is affordable, but it is a new tax on every first contact, and it will be paid by the people with the least to prove it with.

The better answer is a check that does not depend on what the stranger gives you. Look up the real person's own channel yourself, an address or a phone number you found, and ask whether they sent the pitch. Anyone whose name is their business can make that check cheap by keeping one page that says where they write from and where they never do.

Share this

STAY INFORMED

Get AI intelligence like this delivered to your inbox.

Free forever · Unsubscribe anytime


You May Also Find Valuable