How to tell if a LinkedIn profile is fake

To tell if a LinkedIn profile is fake, check its history before its polish. Open the profile's About this member panel (LinkedIn's older name for it is About this profile) to see when it was created and what has been verified, read the work history for details that can be checked, run the photo through a reverse image search, and then look for the person outside LinkedIn, in records other people wrote about them over the years. A real career leaves those traces, and a fake profile usually has the career without them. No single check settles the question, so weigh them together.

LinkedIn, which says it has over 1.3 billion members, reports that between July and December 2025 it stopped 88.9 million fake accounts at registration, restricted 24.4 million more before any member reported them, and restricted 361,300 after members did. LinkedIn says 99.7 percent of the fake accounts it stopped were stopped before a member report.

The fakes that reach you are the ones those defences missed. They come in three kinds, and each needs slightly different checks.

An invented person. NPR reported in March 2022 that Renée DiResta and Josh Goldstein of the Stanford Internet Observatory had found more than 1,000 LinkedIn profiles using what appear to be AI-generated faces. More than 70 businesses were listed as employers. Several told NPR they had hired outside marketers to help with sales, but said they had not authorised computer-generated images.

A real person, impersonated. The FBI warned in 2024 that North Korean cyber actors impersonate people including "general recruiters on professional networking websites", using "pictures stolen from open social media profiles of the impersonated individual."

A real person with an invented history. The name and face are genuine, and some of the jobs, titles or degrees are not. The checks in our guide to verifying a resume apply here.

On the profile, click More and choose About this member. LinkedIn introduced the panel in 2022, as About this profile, to "show you when a profile was created and last updated", and it is also where verifications appear: LinkedIn's help pages say a workplace verification may be visible "in the About this member section".

A creation date tells you how long the profile has existed, and nothing about the career it describes. A profile opened this year that claims fifteen years of work is not evidence of fraud, since people join late or delete an old account and start again, but it cannot vouch for its own history. A profile opened years ago, with activity spread across those years, carries some of its own evidence.

Read any recent change the same way. An old profile that recently got a new photo, headline and location may belong to someone who changed jobs, or may be an account that changed hands.

Yes, although a badge narrows what can be fake. Two of LinkedIn's verifications matter here, and both show as a badge beside the member's name to anyone logged in:

• Identity verification means "a member's government-issued ID was verified by one of LinkedIn's verification partners", and the name used "must match the name on your LinkedIn profile."
• Workplace verification means the member "has confirmed their association with the specific company", through a work email, Microsoft Entra Verified ID, or a LinkedIn Learning or LinkedIn Recruiter licence provided by the company. A work email must match "your current company's email domain."

In September 2025, LinkedIn told reporters it would require workplace verification when a member adds or updates a recruiter title, or a senior title such as vice president, Help Net Security reported, quoting LinkedIn's vice president of trust product. Members who already held those titles were not asked to verify straight away, and LinkedIn's help pages say only that members "may need to complete workplace verification to add or edit your work experience."

LinkedIn is also testing a third kind, announced in September 2026: it asks verified members to confirm the work or school experience of connections they worked or studied with, and "Once someone receives enough confirmations, LinkedIn adds a verification badge". Most profiles will not show it yet. The same announcement lets companies that have verified their LinkedIn Page remove impostors from search results under the company's name.

So an identity badge shows that a verification partner checked a government ID and that LinkedIn matched it to the profile's name, or to an alias it could confirm, not that the jobs, schools or photo are real. A workplace badge shows that someone with access to the company's email, its sign-in or its LinkedIn licences confirmed the association. Neither shows who is using the account today. A fake can carry a badge when a real, verified person invents the rest of their history, or when someone else ends up controlling an account its original owner verified.

A missing badge proves little. Verifications are optional, "identity verification is limited to select geographies", and the work email route is open only to some companies. The exception is a recruiter or executive title added since September 2025. A missing workplace badge there is worth a question, though not proof, since an unusual title may fall outside the rule.

Read the profile as a set of claims, the way you would read a resume. LinkedIn's own guidance on fake profiles points to profiles with little or no information, and to "Job titles that don't make sense, unrealistic career timelines or inconsistent dates."

The profiles the Stanford researchers found had patterns in common. Many used titles like business development manager. They often listed two or three former employers, sometimes well-known names, with no details, and none of the companies NPR asked had records of the supposed employees. NPR contacted 28 universities about 57 of the profiles, and of the 21 that responded, none had records of the supposed graduates. Some profiles claimed undergraduate business degrees from schools that do not offer one.

So check jobs against the employer, and a degree against the school, starting with whether the school offers that degree at all.

Connections count for who they are. A profile shows "500+" once a member has more than 500 connections, so the number says little. LinkedIn's scam guidance says to pay close attention to profiles that "are not connected to many employees within the company they represent." Mutual connections prove less than they seem, because many people accept invitations from strangers.

Activity is the third layer. Posts and comments that other people replied to, spread over years, are hard to produce in a hurry. Endorsements take one click to give, so they count for little either way.

By eye, usually you cannot. In a 2022 study in PNAS, Sophie Nightingale and Hany Farid asked 315 people to sort AI-synthesised faces from real ones, and their average accuracy was 48.2 percent, close to chance. With training and feedback after each face, another 219 people reached 59.0 percent, and a third group rated the synthetic faces 7.7 percent more trustworthy than the real ones. Some synthetic faces, the authors suggested, "simply do not contain perceptually detectable artifacts."

The tells that circulate come from older generators. DiResta's investigation began with a photo in which the woman wore only one earring, bits of her hair disappeared and reappeared, and her eyes sat right in the middle of the image. That last tell has a cause. A 2023 paper by LinkedIn researchers and Farid explains that faces from StyleGAN generators "are aligned in terms of ocular position and interocular distance", and are mostly generated from the neck up, while real profile photos tend to show more of the upper body and shoulders.

Newer generators work differently. In the same 2023 paper, a detector trained on StyleGAN faces caught 0.9 percent of Stable Diffusion faces at a 1 percent false positive rate, because the diffusion process "does not rely on the same type of training from aligned faces." A 2024 LinkedIn model trained on several generators caught 98 percent of AI faces from engines it had been trained on, falling to 84.5 percent for engines it had not seen, with its threshold set to wrongly flag 0.5 percent of real photos. A person looking at one photo has far less to go on, so treat a visual tell as a reason to keep checking, never as a finding.

A reverse image search is more useful, because its results can be checked. Google's search by image can return "Websites with the image or a similar image", and there are three outcomes:

• The photo appears elsewhere under a different name, on pages older than the profile. This is strong evidence that the profile took someone else's photo, the method the FBI describes, and it breaks LinkedIn's rules: "Do not use an image of someone else, or any other image that is not your likeness, for your profile photo." If the other pages are newer than the profile, the copying may run the other way, and a different name can still be the same person under a former or transliterated name, so check before concluding.
• The photo appears elsewhere under the same name, on pages older than the profile. That supports the profile.
• Nothing matches. That fits a generated face and a real person whose photo is not posted anywhere else equally well, so it settles nothing.

Of the checks you can run without contacting anyone, this one carries the most weight. When the Stanford researchers searched for the personas behind the profiles they had found, Goldstein told NPR, "we didn't find any evidence of them in other places, which is rare."

Look for records that someone other than the person wrote, at the time:

• The employer's own website: team pages, blog bylines, press releases, and archived copies of those pages from the years the profile claims.
• Code. GitHub's API returns public information about an account, including when it was created, and pull requests merged into real projects show work other people reviewed.
• Talks, papers and patents, which carry their own dates and affiliations.
• News, conference programmes, alumni announcements, and old forum or social accounts under the same name or handle.
• If the person emailed you, the address itself. Our guide to investigating an email address covers what one can show.

Common names collide constantly, so a match needs two independent details to line up, such as a name and an employer at the same date. Our guide to running an OSINT investigation covers keeping those links honest.

A real professional's footprint builds up sideways. A colleague tags them in a team photo one year, a meetup lists them as a speaker another, and an employer's archived team page shows them one year and not the year before. Other people wrote most of it, across years and across services that have nothing to do with each other.

A fabricated profile tends to be complete, recent and self-contained, written by its author inside a narrow window. A fake profile can copy a career. It cannot easily copy the years of other people mentioning it. Our guide to spotting fake candidates calls this thinness.

If you are hiring, the profile is one more claim from the candidate. Check that employers, titles and dates agree with the resume and any public work, and whether the profile existed before this job search. A thin profile is a reason to ask questions.

If a recruiter has messaged you, find the company's careers page yourself and read how it says recruiters contact candidates, then check whether the recruiter shows a workplace badge and is connected to colleagues there. The warning signs in the FBI's 2024 alert include offers from prominent technology firms that are unexpected or involve "unrealistically high compensation without negotiation", "Requests to move professional conversations to other messaging platforms", and requests for a "pre-employment test" that means running unknown packages, scripts or GitHub repositories. LinkedIn's scam guidance adds, "Do not provide payment or account credentials as part of the application process." The FBI's advice is to verify a contact "using separate unconnected communication platforms."

Recruitment approaches are also used for espionage. In November 2025, the UK's security minister told Parliament that, according to MI5, Chinese intelligence officers were trying to cultivate people with access to sensitive information about Parliament and the government, "often masked through the use of cover companies or external headhunters", and that the National Protective Security Authority would strengthen its engagement with professional networking sites. Job fraud is common in its own right: the FBI's Internet Crime Complaint Center logged 24,688 complaints of employment fraud in 2025, with reported losses of almost $363 million.

If a stranger pitches you a product, the persona may be fronting a real company. NPR found that many of the AI-faced profiles seemed to exist to drum up sales, connecting anyone who replied to a real salesperson. Deal with the company through its own website.

If you review reported accounts, you often meet LinkedIn profiles linked from a seller's storefront or a contractor's application. Compare the reported profiles with each other for the same few employers, the same headline formula, creation dates in the same weeks and the same style of photo. Our guide to linking related accounts covers how much each shared detail is worth. Report the profiles to LinkedIn, and act on your own platform on your own evidence.

The people, company and domain below are invented, and halvermoor.example uses a domain name reserved for documentation.

An engineer at a fintech gets a LinkedIn message from Mara Ellingsworth, Senior Technical Recruiter at Halvermoor Payments, about a staff role at pay well above the market. After two messages, Mara suggests moving to Telegram, then sends a GitHub repository with a take-home exercise to run on the engineer's laptop. The engineer forwards it all to the security team.

The profile. About this member shows it was created in June 2026, with no verifications. LinkedIn said in 2025 that it would ask for workplace verification when a recruiter title is added, so the missing badge is a question worth asking. Three earlier recruiting jobs at well-known companies have one line each. It shows 500+ connections, and the ones the team can see are mostly engineers at fintechs. Its only activity is a few reshared job posts from the past month.

The photo. A clean headshot, eyes centred, cropped at the neck. A reverse image search finds no matches, which on its own settles nothing.

The company. Halvermoor's careers page says its recruiters write only from halvermoor.example addresses and never ask candidates to install software. Its LinkedIn page lists four recruiters, each with a workplace badge, connected to each other and to dozens of colleagues, and posting Halvermoor's job ads for years. None of them shares a connection with Mara.

Outside LinkedIn. The name returns nothing at all.

The confirmation. The security team writes to the recruiting address on Halvermoor's careers page, and Halvermoor replies that it has no recruiter by that name.

The requests match the FBI's warning signs, so nobody runs the repository. The team warns staff, and the engineer reports the profile to LinkedIn as not a real person. The decision rests on the company's answer and on what Mara asked for. Everything else only told the team where to look. Two questions stay open, and the team's notes say so: whether the photo was generated or taken from someone, and who is behind the account.

An aged account. A profile opened years ahead and used lightly gets a creation date that looks earned, though the years are on the profile and not around it.

A bought or taken-over account. A real account comes with real connections, real history and sometimes a badge its first owner earned. The signs are change: a new photo, headline, location or language on an old profile, often after a long silence.

Impersonation of a real person. The photo, the name and often the employer are real, so the outside checks find a real person without proving that they run this profile. Contact the person or their employer through a channel you found yourself, and look for a duplicate of the person's real profile, which LinkedIn lists among its warning signs.

A real, verified person with an invented history. The identity badge is genuine, so only the history checks catch it.

A face from a newer generator, never posted anywhere, leaves the photo checks silent.

These checks raise the cost of a convincing fake without making one impossible. The ones that hold up best go around the profile: the employer's own channel, and records other people wrote.

Most thin profiles belong to real people. New graduates have short histories, and plenty of professionals joined late, rarely post, or deleted an old account and started again. Verification is optional and not offered everywhere. Some people leave the photo off for privacy, work under a name that differs from the one on their documents, or have a name transliterated differently in different countries. Agency recruiters and contractors may not be connected to the staff of the companies they work with. And a real person may have used an AI tool to make a headshot of themselves, which is still their likeness.

So keep three levels apart. Thin means look further. A contradiction, such as an employer or school with no record, a photo under someone else's name, or a company that denies the person, calls for an explanation. A confirmation through a channel you found yourself settles the question. If the profile belongs to a candidate or a customer, ask them about what does not fit before deciding anything.

On desktop, open the profile, click More, then Report / Block, and select Report [name] or entire account. Select This person is impersonating someone or This account is not a real person, then submit the report. LinkedIn says reported members "won't know who made the report." If the profile is impersonating someone you know, tell them through a channel you already have.

Checking a profile by hand means working through company pages, archives, code, talks and news, and noting where every finding came from. Sixtyfour's agent starts from what you already have, such as a name, an email address or a known profile, researches public sources, and links what it finds to the same person. Every finding comes with its source, and anything it could not establish is reported as open rather than filled in. A person on your team decides what to ask and what to do next. There is more on how hiring teams use it on our workforce identity page, and on how platform teams use it on our trust and safety page.

Check when the profile was created and what its verifications actually prove, read the work history for details that can be checked, and search the photo. Then look for the person outside LinkedIn, in records other people wrote over the years, and confirm anything that matters through a channel you found yourself.