LinkedIn Scans 6,222 Extensions. Gem Lost a Page; You Could Lose a Seat

As of October 13, the LinkedIn company pages for Pin, Gem and Neuroscale return "Page not found," and LinkedIn has given no details about why. If your team runs any of these tools, or anything like them, the vendor's survival is the wrong thing to worry about. The question is whether your own recruiter account now carries risk you never agreed to take on.

What actually happened to Gem, Pin and Neuroscale?

Between late September and mid-October, LinkedIn removed the company pages of three recruiting-tech vendors and said nothing about what any of them did. That silence matters. A missing company page tells you LinkedIn objected to something. It doesn't tell you what.

Here is the sequence ERE reported:

• Over a month before Oct 13: Pin CEO Steve Lu's personal LinkedIn profile disappears.
• Late September: Gem's company page is gone.
• Early October: Neuroscale, which launched its Arbi recruiting platform in July, loses its page. Co-founder Ishan Jadhwani had warned other recruiters about Pin's ban shortly before this happened.
• October 13: All three pages return "Page not found."

The vendors responded in different ways. Gem said, "Our customers' partnerships with Gem and/or LinkedIn remain unaffected." Lu went on the offensive with a video: "LinkedIn doesn't get to determine who's in business or not."

Neither statement should settle the question for you. Calling this a ban on the vendors' products goes further than anything confirmed, and Gem's reassurance comes from the vendor, not from LinkedIn. The useful thing to do is look at what LinkedIn can actually detect, and on whose account.

Why is your LinkedIn account at risk, not the vendor's?

The detection runs in your browser and sends what it finds out under your login. The vendor has a legal team and a company page it can lose. You have your own logged-in account with your history and live pipeline attached, and that's the account LinkedIn's code is looking at.

Fairlinked e.V., the group that published its research as "BrowserGate" earlier this year, found that every time someone loads LinkedIn in a Chrome-based browser, a hidden script:

• checks the browser for more than 6,000 installed extensions
• collects 48 hardware and software characteristics about the device
• encrypts the result into a fingerprint
• attaches that fingerprint to every API request made during the session

6,222

LinkedIn confirmed that it scans for extensions and says it does so to identify users who violate its terms of use. Those terms are broad. The policy bans "crawlers", bots, browser plug-ins or extensions that "scrape, modify the appearance of, or automate activity on LinkedIn's website," and warns that users "risk having their accounts restricted or shut down."

Because the fingerprint goes out with requests from a logged-in session, an installed extension is a signal tied to a specific recruiter, whatever the vendor's legal position is. Fairlinked goes further and alleges that LinkedIn has already sent enforcement threats to users of third-party tools it identified through the scanning. That claim hasn't been confirmed. Still, the mechanism exists and LinkedIn admits it runs.

Which sourcing tools carry the most risk to your account?

In-browser extensions carry the most direct risk to your account. Automated messaging comes next. Bought data carries the least account risk and the most provenance risk. Sort every tool your team pays for into one of these three groups before deciding what to keep.

In-browser extensions

These are the easiest for LinkedIn to see. The script probes for the extension's files, so the extension being installed is the signal, whether or not you used it that day. The policy also covers tools that "modify the appearance" of LinkedIn. On a plain reading, a sidebar that overlays emails or scores on a profile page falls under that, even if you think of it as a lookup tool rather than a scraper.

Automated messaging and cloud bots

These give themselves away through behavior, including sessions routed through cloud proxies shared with other customers. One industry analysis concluded that browser extensions and shared cloud proxies are the setups LinkedIn is best equipped to detect. If a sequence sends from your account, your account takes the restriction.

Bought data with unclear provenance

A purchased contact feed doesn't touch your LinkedIn session, so it poses little direct risk to your account. The risk is upstream. On September 16, 2026, in case 5:25-cv-08393-PCP before Judge P. Casey Pitts, a consent judgment ordered ProAPIs Inc, Netswift (SMC-Private) Limited and Rehmat Alam to stop scraping, stop selling scraped data, stop using fraudulent accounts, and delete everything they had already taken.

According to LinkedIn's complaint, that data came from a rotating pool of over a million counterfeit accounts, and customers paid up to $15,000 a month for it. Earlier, hiQ Labs ended with a $500,000 consent judgment and its own order to delete data. The lesson for buyers: a feed like this can stop overnight, and your CRM may then hold records whose collection you can't defend.

The platform you source on now sells the same thing

LinkedIn is both the referee and a competitor in this market. At its Talent Connect conference on September 29, around the time Gem's page disappeared, LinkedIn announced Hiring Assistant 2, which expands its own agentic recruiting product.

Any workflow built on reading LinkedIn now depends on a platform that sells a competing product.

LinkedIn frames the enforcement as privacy protection, and some of it plausibly is. But the same action also clears out third-party tools that compete with a product LinkedIn has just launched. The practical point isn't whether the motive is good or bad. LinkedIn has every reason to keep narrowing what outside tools can read, so plan your stack as if that access will keep shrinking.

What Refolk's index says about sourcers and LinkedIn Recruiter

Almost no sourcers in Refolk's index list LinkedIn Recruiter as a skill: 143 of the 5,739 people with sourcer titles in the US. That says more about what sourcers advertise than about what they use.

Read the UK figure as an upper bound. "Sourcing Specialist" also picks up procurement jobs, and that shows clearly in the UK sample (Bloom Procurement and Capgemini both appear). The top UK employers include Deloitte and HSBC. On the US side, the group includes sourcers at Meta, MongoDB, Anduril, Tesla and Apple.

2.5%

Here is my reading, which the data doesn't directly measure: sourcers treat a Recruiter seat like a utility, not a skill, so they don't mention it. If that's right, the same blind spot is likely to show up in a tool audit. A team can't see how much of its pipeline depends on one platform until that platform starts removing pages.

How do you keep filling reqs without tools that read LinkedIn?

Find candidates where they publish their own work: public code, personal sites, talks and package registries. Then use LinkedIn by hand, inside your own session, to check the people you've already found.

GitHub is the obvious place to start, with 180 million-plus developers according to Octoverse 2025. It only covers part of the market, though. About 81.5% of 2025 contributions happened in private repositories, so a GitHub-only search misses most engineers whose work happens inside a company. A pipeline that holds up layers several sources:

1. Public commits and maintained repos show what someone builds today and in which language.
2. Package registries name the people who maintain the libraries your team actually depends on.
3. Conference talks surface engineers whose day jobs are hidden in private repos.
4. Personal sites and technical blogs often list employers, stack and a contact route LinkedIn never shows.
5. LinkedIn, checked by hand confirms current title and tenure for a shortlist you built elsewhere.

Doing this by hand is slow. You cross-reference a commit email against a talk abstract against a personal domain, and do it again for every candidate on the list. That stitching is the gap Refolk closes. You describe the person in plain English and get back a ranked shortlist drawn from GitHub, LinkedIn and the open web. It's a search you run, not an extension sitting in the browser where your Recruiter seat is logged in.

A five-step audit of your sourcing stack

Run the audit across every seat, not only your own, because the fingerprint is tied to each recruiter's logged-in session. Five steps cover it:

1. List every extension in every browser profile where a recruiter logs into LinkedIn. Compare the list against the three groups above.
2. Remove anything that scrapes, overlays or automates on LinkedIn from those browsers. LinkedIn's policy names all three behaviors, so removing only the "scrapers" doesn't cover you.
3. Turn off sequences that send from LinkedIn accounts. Keep outbound sequencing on email and other channels you control.
4. Ask data vendors for provenance in writing: where the data was collected, through which accounts, and whether any of it came from LinkedIn. Quarantine any feed whose vendor can't answer.
5. Tag CRM records by source, so that if a vendor gets a ProAPIs-style deletion order, you know which records it covers.

The hardest case is the req that's already in flight. The hiring manager has seen a slate built partly with a tool you've just removed, the next slate is due this week, and rebuilding the top of the funnel from open sources takes time you haven't budgeted. For hard-to-fill roles, this is exactly when an outside sourcing channel is worth paying for.

Once the open-web pipeline is running, it's sturdier than the setup it replaces. No vendor page, extension list or LinkedIn product launch can switch it off.

FAQ

Are Gem, Pin and Neuroscale banned from LinkedIn?

Their company pages are gone, and Pin CEO Steve Lu's personal profile has been removed. LinkedIn hasn't said what it objected to, and Gem says its customers' partnerships with Gem and LinkedIn are unaffected. Treat the removals as a sign of how LinkedIn sees these tools, not as a confirmed ban on using them, and don't assume you're safe either.

Is my LinkedIn account at risk if I use a sourcing extension?

It can be. LinkedIn's terms ban extensions that scrape, change the appearance of, or automate activity on its site, and warn that accounts can be restricted or shut down. Its code checks for thousands of extensions (6,222 per Fairlinked, 6,236 per BleepingComputer) and sends a device fingerprint with your logged-in requests, so the signal is tied to you, not to the vendor.

Does bought contact data put my LinkedIn account at risk?

Not directly, because no LinkedIn session is involved. The risk is provenance. ProAPIs was ordered in September 2026 to delete everything it had scraped, data LinkedIn says came through more than a million counterfeit accounts. Buyers of that kind of feed can lose the supply overnight and be left holding records they can't vouch for.

Can GitHub replace LinkedIn for technical sourcing?

Only partly. GitHub has 180 million-plus developers, but about 81.5% of 2025 contributions were in private repositories, so most engineers at large companies don't show up in public commits. Combine GitHub with package registries, conference talks and personal sites, then use LinkedIn by hand to check your shortlist.