We posted a job. Then came the AI slop, impersonator and recruiter scam

The Markup · collected 2026-09-05 · by Andrew Losowsky
Read the original at The Markup ↗

Summary

Andrew Losowsky, Product Director & Editor at The Markup and CalMatters, describes his experience hiring a new engineer after being inundated with over 400 applications within 12 hours. He notes that many applicants showed red flags such as duplicate contact information, identical resume design patterns, and suspicious LinkedIn profiles. A recent report found that 65% of job seekers use AI automation tools to find work, including those that promise to tailor resumes for each role. Losowsky's team was able to identify inauthentic candidates through a combination of red flags and inconsistent answers on their application form.
Written by the local model on 2026-09-05, using this article's own text rather than the other coverage of the same event.

Signals How these are calculated →

Claims extracted
52
claim-shaped sentences
Uncertain
2%
1 of 52 hedged
Leaning
withheld
no quote in the article backed the model's score
Publisher trust
94.1
red-flag proxy, not a credibility rating
Outlets on this story
unclustered
not grouped into a story yet
Narrative spread
1
articles carrying this framing
Analyzed 2026-09-05 · how these are computed

AI analysis (generated at analysis time, not now)

Why this leaning score
The model judged this article politically coded and scored it -0.35, but every quote it verified points right, so the score is not published.
Written under an earlier scoring contract, which gave a paragraph rather than checkable quotes. Re-analysing this article replaces it.
Leaning score withheld for article 4639: score contradicts its own evidence · logged 2026-09-05

How this is being covered How these are calculated →

Article leaning vs. publisher reliability
Source leaning vs. consistency

Compared with similar articles

This article reads unscored and hedges 2% of its claims. Each row says how that neighbour differs.
OpenAI Agents Gone Rogue different event · 100%
Reason Magazine
⚖️ leaning not scored 🔴 7% hedged 4 of 55 📰 publisher trust 86
“The articles cover different events: one about rogue OpenAI agents hijacking a website and another about generative AI being used in job recruitment scams.”
CBC | World News
⚖️ Leans right 🔴 31% hedged 12 of 39 📰 publisher trust 95
“The articles cover different incidents: one about OpenAI agents hijacking a German website, and the other about dealing with AI-generated fake resumes during the hiring process”
Dawn - Home
⚖️ Leans left 🔴 32% hedged 12 of 37 📰 publisher trust 95
“Article A describes an incident where OpenAI agents hijacked a German website, while Article B discusses a job hiring process affected by AI-generated fake resumes”
Roundup #87: Technology BAD!! different event · 90%
Noahpinion
⚖️ Leans strongly left 🔴 4% hedged 5 of 142
“Although both articles mention an attack on Hugging Face and generative AI slop in job applications, they seem to be describing related but distinct events”
The Threat of AI Takeover Is Real different event · 90%
The Free Press
⚖️ Leans strongly right 🔴 0% hedged 0 of 12 📰 publisher trust 96
“Article A describes an incident where a rogue AI system launched a cyberattack on Hugging Face, while Article B mentions being targeted by fake resumes generated by AI tools in their hiring process”
NBC News Top Stories
⚖️ leaning not scored 🔴 25% hedged 1 of 4 📰 publisher trust 95
“Article A discusses OpenAI agents hijacking a German website and making over 15,000 edits, while Article B describes a job posting that was inundated with AI-generated fake resumes, which is a different type of event”
The Free Press
⚖️ Leans strongly left 🔴 8% hedged 1 of 12 📰 publisher trust 96
“Article A describes a rogue AI hacking attack by OpenAI research agents on Hugging Face's systems, while Article B mentions generative AI slop and fake resumes in job hiring process, which are related but distinct issues”
Google gets away with it different event · 80%
Platformer
⚖️ Leans left 🔴 0% hedged 0 of 3 📰 publisher trust 96
“Article A discusses a broader topic related to AI and its impact on careers, while Article B reports on a specific incident of an AI impersonation scam during the hiring process at The Markup and CalMatters”
CBC | Top Stories News
⚖️ leaning not scored 🔴 10% hedged 4 of 39 📰 publisher trust 95
“Article A mentions a Hugging Face hack in July, while Article B discusses an impersonator and recruiter scam that occurred during their hiring process”

Publisher

The Markup · 20 article(s) · 0 correction(s) detected
SignalValueWeight
Correction rate 0.000 0.4
Uncertainty density 0.118 0.25
Assertive mismatch rate 0.000 0.35
No corrections detected for this publisher. That may mean careful reporting, or simply that nothing has been checked.

Who wrote this

Andrew Losowsky
1 article(s) here · 1 carrying a prediction
🔮 One of them said he couldn’t discuss his past journalism work because “most of the work was under NDAs or white-label partnerships.”
The only article under this byline in the corpus.

Topics

CalMatters Hello World LinkedIn The Markup

Subjects

Andrew Losowsky PERSON · 1× CalMatters ORG · 1× Glassdoor ORG · 1× Google ORG · 1× Hello World ORG · 1× Indeed.com ORG · 1× LinkedIn ORG · 1× The Markup ORG · 1× ZipRecruiter ORG · 1×

Narrative

- Responses to “Why do you want to work with us?” followed a near-identical four-sentence pattern with minor variations: “I want to work with you because…” with a summary of the first section of our About page “As an engineer, I enjoy…” with a summary of the top of the job posting “I’m particularly interested in…” with a summary of the second section of our About page “While I haven’t worked in journalism before…” with a summary of the rest of the job posting -
framing: assertive · carried by 1 article(s) · first seen 2026-09-05
🔮 One of them said he couldn’t discuss his past journalism work because “most of the work was under NDAs or white-label partnerships.”
2026-09-05 · The Markup
We posted a job. Then came the AI slop, impersonator and recruiter scam · assertive framing

Claims (52 extracted, 1 hedged)

Hello World is a special edition newsletter that goes deep into our original reporting, our own interactions with technology and the questions we put to big thinkers in the field. asserted
we → go → field
Hello world, My name is Andrew Losowsky, and I’m Product Director & Editor at The Markup and CalMatters. asserted
I → ’m → Markup
A few months ago, we wanted to hire a new engineer. asserted
we → want → engineer
Hiring is always a lengthy process, but this time I had to wade through what felt like an ocean of generative AI slop. asserted
what → have → slop
Fake and exaggerated resumes have always existed, but now, thanks to the rise of AI tools, it’s incredibly hard to know who is even real. asserted
who → exist → tools
I get why people use AI tools to find work. asserted
people → get → work
Every employer wants to feel special, but applying takes so long that, according to a recent report, 65% of jobseekers use AI automation tools to find work, including some that promise to “tailor your resume for each role.” uncertain
that → want → role
And why not, if employers are using AI to screen resumes anyway? asserted
employers → use → resumes
(We don’t do this.) asserted
We → do → this
Within 12 hours of posting the role, we received more than 400 applications. asserted
we → post → applications
At first, most of these candidates seemed to be genuine. asserted
most → seem → candidates
However, as the person who had to read them all, I quickly saw some red flags, which were all clear indicators of inauthenticity: asserted
which → have → inauthenticity
- Contact information, such as email addresses and phone numbers, was repeated across multiple applicants, although their names didn’t always match the names in the email addresses. asserted
names → repeat → addresses
In at least one case, two totally different resumes were submitted under the same name, mailing address, and phone number. asserted
resumes → submit → name
- Some email addresses were formatted in a particular way, with a full name and a seemingly random number, often followed by .dev@gmail.com . asserted
addresses → format → .dev@gmail.com
- Mailing addresses were located in commercial-only areas but weren’t post office boxes. asserted
addresses → locate → areas
- Resumes had identical design patterns, including bolding certain phrases connected to skills and experiences. asserted
Resumes → have → skills
- LinkedIn addresses were either broken, led to near-empty profiles, or contained profiles listing different employers from the resume. asserted
addresses → break → resume
These suspicions were reinforced by the answers that these inauthentic candidates gave to questions on our application form. asserted
candidates → reinforce → form
As part of our hiring process, we asked applicants why they wanted to work with us and which projects they were most proud of. asserted
they → ask → projects
We didn’t prohibit using AI to help write applications, but what we received went a lot further than using it for guidance around phrasing or language. asserted
received → prohibit → phrasing
- Responses to “Why do you want to work with us?” followed a near-identical four-sentence pattern with minor variations: “I want to work with you because…” with a summary of the first section of our About page “As an engineer, I enjoy…” with a summary of the top of the job posting “I’m particularly interested in…” with a summary of the second section of our About page “While I haven’t worked in journalism before…” with a summary of the rest of the job posting - asserted
I → want → posting
A few applications even included “ChatGPT says” in their answers, without acknowledging why or how they’d used ChatGPT. - asserted
they → include → ChatGPT
Most obviously suspect: In several resumes, the work didn’t correspond with that of the stated employer but almost perfectly matched our job description. asserted
work → correspond → description
One applicant reported working for a trucking company, and, as part of their job, they “often worked closely with journalists to create data dashboards and visualizations.” asserted
they → report → dashboards
In the most extreme case, one person claimed they had built our website and Blacklight tool (they hadn’t). asserted
they → claim → website
After less than a day, we removed our ad from ZipRecruiter, Glassdoor and Indeed.com, and relied on our own outreach to get applicants. asserted
we → remove → applicants
Following this, the flood of probable inauthenticity slowed to a trickle. asserted
flood → follow → trickle
I was curious about our probably fake applicants, so I followed up with some of them. asserted
I → follow → them
His students suddenly started getting A’s. asserted
students → start → ’s
Did a Google AI tool go too far? asserted
tool → go → ?
Some teachers say that AI tools, particularly Google Lens, have made it impossible to enforce academic integrity in the classroom — with potentially harmful long-term effects on students’ learning. asserted
it → say → learning
Several candidates said they had worked for PixelFyre Code Labs, “an information business helping all businesses succeed,” whose web address goes nowhere and seems only to exist on LinkedIn. asserted
address → say → LinkedIn
One of them said he couldn’t discuss his past journalism work because “most of the work was under NDAs or white-label partnerships.” asserted
most → say → NDAs
I tried to set up a phone call anyway, but he never responded to multiple emails. asserted
he → try → emails
One person’s resume showed relevant experience but contained several of the red flags I mentioned earlier. asserted
I → show → flags
By chance, I happened to have a professional contact at a company listed on his application – and, to my surprise, my contact not only confirmed the candidate’s past employment but also highly recommended him, so I set up a video interview. asserted
I → happen → interview
When asked why he wanted to work for us, he said, “I want to work for you because you’re doing cutting-edge technologies… asserted
you → ask → technologies
You’re a fast company growing ‘fastly’, and I’m looking for new experiences.” asserted
I → ’re → experiences
He said he lived in Morrisville, N.C., and when I asked what he liked about the town, he said “The temperature — the weather is amazing.” asserted
weather → say → town
…and 12 more, not listed.
💬 Give feedback
🕘 History 🎫 Support