Refolk
November 18, 2026·10 min read

Pin Got Its LinkedIn Page Back. 1,863 Recruiters Still Carry the Risk

LinkedIn removed Pin, Gem, and Neuroscale's pages, then restored Pin's after its docs changed. Here's what sourcing teams can still safely automate.

pin linkedin bannedgem linkedin page removedlinkedin automation recruiting tools banai sourcing tool linkedin compliancelinkedin recruiter alternative
Pin Got Its LinkedIn Page Back. 1,863 Recruiters Still Carry the Risk

If an AI sourcing tool sends LinkedIn messages for you, the account at risk is yours, not the vendor's. Between September and October 2026, LinkedIn removed the company pages of Pin, Gem, and Neuroscale. It brought Pin's back only after Pin's docs stopped describing automated LinkedIn sends.

Pin says LinkedIn restored its company page in November 2026 "after reviewing how Pin works." By then, Pin's product copy had changed. LinkedIn connection requests and messages are now described as tasks the recruiter sends from their own account. If you're evaluating an AI sourcing agent, that change shows you where the line sits. It also shows you whose LinkedIn seat is exposed when a vendor gets it wrong.

What happened to Pin, Gem, and Neuroscale on LinkedIn?

In roughly four weeks, LinkedIn removed the company pages of three AI recruiting vendors, along with a CEO's profile. LinkedIn has never named the specific policy involved.

Here's the sequence, pieced together from vendor posts and ERE's reporting:

  1. Sep 11, 2026: Pin publishes its first "Quick note" saying "LinkedIn's team raised concerns about our presence on their platform." The page was already gone by then. Noon cites a Pin post from Sept 12.
  2. Sep 16, 2026: Pin CEO Steven Lu posts a longer follow-up. It claims "our 2,000+ recruiting teams are unaffected."
  3. Sep 29, 2026: LinkedIn announces Hiring Assistant 2 at Talent Connect, presented by VP of Product Dan Reid.
  4. Late September: ERE reports that Gem's company page is gone.
  5. Early October: Neuroscale, which launched its Arbi recruiting platform in July, loses its page. Its CEO, Ishan Jadhwani, had posted a "PSA" that Pin was banned weeks earlier.
  6. November 2026: Pin's homepage FAQ says the page has been restored.

David Manaster at ERE and recruiter Brano Vargic both documented the removals as they happened. LinkedIn gave ERE one statement on the record: "LinkedIn takes action when necessary to prevent our members' information from being scraped and used without their consent." Gem was the only one of the three to give ERE a written statement: "Our customers' partnerships with Gem and/or LinkedIn remain unaffected."

The public record is thin: one sentence from LinkedIn, two reassurances from vendors, and no named policy.

What changed before Pin's page came back?

Pin's product copy stopped promising automated LinkedIn sends, and then LinkedIn restored the page. The order of events is clear. The cause isn't, so read it as correlation and not as a confirmed deal.

These are the observable changes:

  • Pricing page: according to Noon, the automatic LinkedIn connection-request language was gone from Pin's pricing page by September 2026.
  • Reviews page: Pin now says "LinkedIn touches in a sequence are tasks the recruiter sends from their own account."
  • Founder messaging: after the restoration, Lu's LinkedIn feed promoted an agent that can "source and approve candidates automatically" and "reach out to candidates with fully customized emails."

The third point matters most. The automation didn't go away. It moved off LinkedIn and into email, where LinkedIn isn't the one setting the rules. Pin still automates sourcing, candidate approval, and outreach. The only step it handed back to a person is the LinkedIn send.

That's the practical answer to the LinkedIn automation ban question. LinkedIn didn't object to AI in recruiting, since it sells AI recruiting itself. What it objects to is software acting on LinkedIn on a member's behalf.

Who actually carries the risk: the vendor or the recruiter?

The recruiter whose account sends the message carries the risk. LinkedIn penalized the vendors' brands by removing pages and a CEO profile, but its Help Center policy puts account restriction on the member who uses the prohibited tool.

The policy bars third-party software "including 'crawlers', bots, browser plug-ins, or browser extensions that scrape, modify the appearance of, or automate activity on LinkedIn's website." A vendor can lose a company page and keep selling. A recruiter who loses their LinkedIn account loses access to the network and pipeline they built on it.

1,863
US recruiters, technical recruiters, TA partners, and sourcers who list "LinkedIn Recruiter" as a skill
In Refolk's index, these are the people whose own accounts sit on the send side of any automated LinkedIn sequence.

This figure counts people who list the skill, not licensed seats. Still, it shows who's exposed. When Pin and Gem each told the market that customers were "unaffected," they were talking about the vendor relationship. They couldn't vouch for any individual recruiter's account, because the vendor never owned that account.

Why are generalist recruiters the most exposed?

Most LinkedIn outreach is sent by busy generalist recruiters, not by dedicated sourcers. Generalists are also the people most tempted to let a tool click Send for them.

In Refolk's index, 4,994 people in the US hold dedicated sourcer titles: Technical Sourcer, Talent Sourcer, Sourcing Specialist, or Senior Sourcer. In a 25-profile sample of US recruiters who list the LinkedIn Recruiter skill, the top title was Technical Recruiter (11 of 25), not sourcer.

SegmentCountSource
US, dedicated sourcer titles4,994Refolk's index
UK, dedicated sourcer titles431Refolk's index
US-to-UK sourcer multiple~11.6xDerived
US recruiter/TA titles listing "LinkedIn Recruiter" skill1,863Refolk's index
Pin's claimed customer base2,000+ recruiting teamsPin (vendor claim)
Companies on LinkedIn's agentic hiring products20,000+HeroHunt

One caveat on the UK number. "Sourcing Specialist" also matches procurement jobs, and about 5 of 25 sampled UK profiles were procurement, billing, or tutoring roles, compared with about 2 of 25 in the US. That inflates the UK count, so the 11.6x gap is probably larger in reality.

Here's the mechanism. A technical recruiter carrying a full req load with no sourcing support has no time to hand-write connection notes all week. A tool that promises to send them automatically fixes that week. It also puts the recruiter's own account at risk, and nobody on the team is set up to notice until the account gets restricted. When sourcing for a hard-to-fill role already takes too long, the shortcut feels free.

The way out isn't to do everything by hand. It's to automate the parts LinkedIn doesn't police and save human time for the one step it does.

Where is the line, and what can you safely automate?

The line is whether a person clicks Send on LinkedIn. Search, enrichment, scoring, and off-LinkedIn outreach can be automated. LinkedIn connection requests and messages should go out from a person's own account, one deliberate click at a time.

Safe to automate

  • Search across GitHub and the open web. Commit history, conference talks, personal sites, and papers are public and outside LinkedIn's terms.
  • Enrichment. Matching a GitHub handle to a work email or personal site doesn't touch LinkedIn's website.
  • Scoring and ranking. Let the agent decide who's worth contacting.
  • Email sequences. This is where Pin moved its own automation.
  • Drafting LinkedIn messages. A tool can write the note. Writing isn't sending.

Keep under human control

  • Connection requests and InMails. These should be queued as tasks and sent by the account owner.
  • Profile visits at volume. Browser extensions that load profiles for you fall under "automate activity on LinkedIn's website."
  • Anything that scrapes LinkedIn pages. That's the exact behavior in LinkedIn's one on-record statement.

This is where Refolk fits. It does the search and enrichment half across GitHub and the open web, from a plain-English description of the person you want. You get back a ranked shortlist with the context you need to write a good note, and you decide who gets a LinkedIn message and when.

The result is a short daily queue of well-chosen LinkedIn sends, each one approved by a person, from an account that stays yours.

Why don't "compliant" automation claims mean what buyers think?

When a vendor says its automation is "compliant," it's usually describing how hard the tool is to detect, not whether LinkedIn's policy allows it. LinkedIn judges a tool by how it accesses the site, not by whether it calls itself automation.

One analysis from Tactera Digital puts it plainly: LinkedIn "draws its line by access method rather than by the word 'automation.'" A cloud-based sender with randomized delays and a per-day cap is still software acting on LinkedIn for a member. The cap is a ceiling the vendor's software enforces. LinkedIn never approved it.

A daily send quota is a ceiling the vendor allows, not permission LinkedIn granted.

So when you assess an AI sourcing tool for LinkedIn compliance, the useful question isn't "is this safe?" Ask what the tool does on LinkedIn's website, and whose session it uses to do it. If the answer is "it sends from your account while you sleep," Pin's rewritten docs show which side of that line a vendor chose before its page came back.

Is LinkedIn acting as referee or competitor?

LinkedIn is both: it polices third-party tools on its site and sells its own agentic recruiting product that does the same front-end work. The timing deserves a look. LinkedIn launched Hiring Assistant 2 on Sept 29, Gem's page was gone by late September, and Neuroscale's went down in early October.

HR Briefing reports that LinkedIn's agentic hiring business has passed $450M in annualized revenue. HeroHunt puts it at more than 20,000 companies on LinkedIn's agentic hiring products.

$450M+
Annualized revenue reported for LinkedIn's agentic hiring business
The first Hiring Assistant mostly handled intake, sourcing, and outreach, which is the same work third-party agents automate.

The first Hiring Assistant "mostly handled the front end of recruiting: intake, sourcing and outreach." A third-party tool that automates LinkedIn sends competes with that product directly. That doesn't prove the removals were about competition. LinkedIn's stated reason is member data and consent, and that reason is enough on its own. But it does mean you shouldn't expect LinkedIn to be generous about automated sends from outside vendors. If you're looking for a LinkedIn Recruiter alternative, the safest one doesn't try to copy Recruiter inside LinkedIn. It finds people somewhere else.

For technical roles, much of the signal was never on LinkedIn to begin with: who merged what, who maintains which library, who gave which talk. Searching that layer directly, which is what Refolk does, surfaces candidates whose LinkedIn profiles say little, without automating a single action on LinkedIn.

What should you ask an AI sourcing vendor now?

Ask what the tool does on LinkedIn's website, whose account it does it from, and what the vendor's docs say today. Pin's copy changed within a few months, and so can anyone's.

Use this checklist in your next demo:

  1. Does any part of the product send connection requests or messages from my LinkedIn account without my click? The right answer after November 2026 is no.
  2. Does it use a browser extension or cloud session that loads LinkedIn pages for me? If yes, that's the "automate activity" clause.
  3. Where does the profile data come from? "We scrape LinkedIn" is the behavior in LinkedIn's statement. Public GitHub and web sources aren't.
  4. If LinkedIn restricts my account, what does the vendor do? Pin and Gem both said customers were unaffected at the vendor level. Ask what that means for an individual seat.
  5. Where does the automation live now? Pin moved its automation to email. Ask your vendor to point to the same change in their product.

A good vendor answers all five in writing. A vendor that tells you "we're compliant, we have daily limits" has told you how its tool avoids detection, and nothing about what LinkedIn's policy allows.

FAQ

Was Pin banned from LinkedIn?

Pin's company page was removed in September 2026. Pin's first note about it is dated Sep 11, and CEO Steven Lu's longer follow-up came on Sep 16. Pin says LinkedIn restored the page in November 2026 after reviewing how the product works. LinkedIn has never named the policy involved, and Pin says its 2,000+ recruiting teams were unaffected.

Why was Gem's LinkedIn page removed?

ERE reported Gem's company page was gone by late September 2026, and Neuroscale (Arbi) lost its page in early October. LinkedIn's only on-record explanation is that it acts "to prevent our members' information from being scraped and used without their consent." Gem said customer partnerships with Gem and LinkedIn were unaffected.

Can I still use AI sourcing tools without risking my LinkedIn account?

Yes, as long as the tool doesn't act on LinkedIn for you. Automating search, enrichment, scoring, and email outreach across GitHub and the open web is outside LinkedIn's automation policy. Let the tool draft LinkedIn messages if you like, but send connection requests and InMails yourself from your own account.

Is LinkedIn Hiring Assistant 2 the safe alternative?

It's LinkedIn's own product, so it isn't the third-party automation LinkedIn's policy targets. But it's built around LinkedIn, and for technical roles a lot of the best signal sits on GitHub, in conference talks, and on personal sites. Finding candidates there is a separate search, and one you can automate without touching LinkedIn.

The data behind this

Try it on the search you came here for

Stop building boolean strings. Just describe the person.

Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.

  1. 01Describe them

    One plain sentence. Role, city, stack, stage, whatever matters to you.

  2. 02I read the web live

    GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.

  3. 03You read the shortlist

    Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.

  • No boolean, no filters, no seat to buy. One box.
  • Read at search time, so a profile updated yesterday counts today.
  • Every step visible as it runs, every name with its reason.

500 free credits on sign-up. No card, no demo call. See real searches.

Read next