Fake Recruiters With AI Faces: How to Check Who Is Offering You a Job

What the scam looks like now

The FBI counted 24,688 employment fraud complaints in 2025, up from 15,443 in 2023.¹ The FTC, which collects its own reports, says one in three people who lost money to a job or business opportunity scam last year said it started on social media.² Its April 2026 alert describes the current script. An unsolicited text, WhatsApp or Telegram message arrives from a "recruiter" at a well-known company. The role is vague, something like "online assessor," the pay is mentioned before the work is, and you are asked to reply YES instead of clicking a link.³ The money comes later: you deposit a check and send back the difference, or you pay in to unlock earnings from tasks that were never real. The FTC's rule fits in five words. Never pay to get paid.

Japan has the same pattern under a different name. The National Consumer Affairs Center logged 3,694 consultations about "easy task" side-job schemes in fiscal 2023, up from 1,341 in fiscal 2020. The share that began on social media rose from 23% to 63.4%, and the average amount involved went from about ¥280,000 to about ¥763,000.⁴ In May 2026 the center warned that cases among people in their 20s and 30s were rising fast, including victims told to borrow from consumer lenders and send the money on as crypto.⁵

Why the face is so often generated

In 2022 Renée DiResta of the Stanford Internet Observatory received a LinkedIn sales pitch from a "Keenan Ramsey" whose photo looked wrong. She and her colleague Josh Goldstein, working with NPR, went on to find more than 1,000 LinkedIn profiles with apparently computer-generated faces, listing more than 70 businesses as employers.⁶ Those accounts were doing sales outreach, not fraud, and NPR found no illegal activity. They still show the economics. A generated face costs nothing, belongs to nobody, and has no owner who might one day see it and complain.

It also works on the person looking at it. In a 2022 study in PNAS, Sophie Nightingale and Hany Farid found that participants could not reliably tell StyleGAN2 faces from real ones, and rated the synthetic faces 7.7% more trustworthy on average.⁷ If you want a stranger to keep talking to you about money, that is a useful property.

The third reason matters most for how you check. A scammer who copies a real person's photo can be caught with a reverse image search, because the same face turns up on the real person's profile. A generated face has no original to find. The search comes back empty, and an empty result feels reassuring when it should not.

The 2022 tells are out of date

The Stanford researchers spotted many of those profiles by eye: eyes sitting in exactly the same place in every photo, a single earring, hair melting into the background.⁶ Those were habits of GAN generators, which learned from tightly cropped, aligned faces. LinkedIn's own 2023 research with Farid showed how strong that alignment was. Average 400 GAN profile photos together and you get a sharp face; average 400 real ones and you get a blur. Their detector caught 99.6% of StyleGAN faces while flagging 1% of real profile photos, but the authors reported that it did not generalize to faces made with Stable Diffusion.⁸ A follow-up paper extended the approach to diffusion models and to images as small as 128 by 128 pixels.⁹

The plain reading is that platforms have become fairly good at catching the cheap faces of a few years ago, and the diffusion-made headshots that replaced them are harder. Looking for centred eyes or a missing earring will mostly catch the scammers who never updated their tools. For the rest, see the signs that still hold up, and expect fewer of them each year.

Two checks that answer different questions

A reverse image search and an AI photo check often get treated as the same step. They are not. The search asks whether this photo belongs to someone else. The photo check asks whether the face shows signs of being generated or altered.

Run them in that order. If the search finds the same face under another name, you have your answer, and a detector will add little, since a stolen real photo looks real. If the search finds nothing, that is the case a photo check exists for. Open the full-size image rather than the thumbnail, crop it to the one face, and run it through a tool that scores portraits. Our own service, Maat Scan, scores photos of people from 0 to 100 for signs of AI generation or editing. The score is a statistical reference, not a verdict.

Two cautions keep that score honest. Plenty of real recruiters use AI headshot apps, so a high score is a reason to confirm the person some other way, not proof of a scam. A low score, in turn, says nothing about whether the person behind the profile is who they claim to be. The photo is one piece of evidence. The questions in the last section carry more weight.

What LinkedIn now shows you

LinkedIn started giving recruiters verification badges in April 2024, based on a work email at a verified company or an ID check through partners such as CLEAR.¹⁰ On September 23, 2026 it added more. Verified company pages can remove accounts that falsely claim to work there, members can vouch for a colleague's work history, and some companies are testing a rule that requires workplace verification before anyone can list them as a current employer. LinkedIn says its tools have verified 115 million members and more than 700,000 companies.¹¹

Read the badge for what it says. It confirms an identity or an employer at the time of the check. It says nothing about intentions, and a missing badge is not evidence of fraud, because many genuine recruiters have never set one up. Still, a recruiter who appears on the company page and carries a workplace badge has passed a test that would be expensive for a scammer to fake. That counts for more than any judgment about their face.

Five minutes before you reply

Most job scams fail one of these long before the photo matters.

• How did they reach you? An unexpected text, WhatsApp or Telegram message about a job is the FTC's first warning sign.³
• Is money moving toward them? Fees, equipment, a check to deposit and partly send back, or tasks you have to pay into all mean stop.
• Is the job listed on the company's own careers page? Find it yourself, not through a link they sent.
• Can you reach the recruiter through the company? Use the phone number or email on the company's website, not the one in the message.
• What does the profile show? Its age, its connections and activity, whether it has a workplace badge, and whether it appears on the company page.
• Then the photo: reverse search first, and a photo check if the search comes back empty.

If you have already paid or sent ID documents, call your bank first, then report it: ReportFraud.ftc.gov³ and ic3.gov¹ in the US, or 188 (the consumer hotline) and #9110 (police consultation) in Japan.⁴ The recruiter will be gone by then. The report is how the next person's warning gets written.