LinkedIn Automation Rules: What LinkedIn Has Published

LinkedIn’s rule on automation tools is one sentence in its User Agreement. In section 8.2 every member agrees not to “use bots or other unauthorized automated methods to access the Services, add or download contacts, send or redirect messages”, and LinkedIn publishes no exception for third-party automation tools. Everything else people call LinkedIn’s automation policy is one of three things: a limit LinkedIn publishes, an enforcement process it describes, or an action it took against a named company on a known date.

This page lays out those three from the source, in that order, then checks two claims from the 2026 “crackdown” articles that are now repeated in AI answers. It quotes LinkedIn, the courts and the companies involved, and nobody’s estimate. What to do once an account is restricted is on LinkedIn account restricted, and the prevention steps are on how to avoid getting banned on LinkedIn.

The rule: one clause in the User Agreement

The User Agreement is the contract every member accepts at signup, and its section 8.2 is a list of things members agree not to do. Two lines in that list are the whole of LinkedIn’s written rule on automation and data collection. Members agree not to:

• Automate access (8.2, item 13). “Use bots or other unauthorized automated methods to access the Services, add or download contacts, send or redirect messages, create, comment on, like, share, or re-share posts, or otherwise drive inauthentic engagement.”
• Scrape or copy profiles (8.2, item 2). “Develop, support or use software, devices, scripts, robots or any other means or processes (such as crawlers, browser plugins and add-ons or any other technology) to scrape or copy the Services, including profiles and other data from the Services.”

Both lines are quoted as LinkedIn’s page read on 2 November 2026. The Ninth Circuit quoted section 8.2 as it stood at the time in hiQ Labs, Inc. v. LinkedIn Corp., decided 18 April 2022, and the scraping line has been reworded since, so a quote copied from the opinion is not the current text.

Read the first line for what it does not say. It names the activities, adding or downloading contacts, sending messages and engaging with posts, but no tool, no category of tool and no volume. It turns on the word “unauthorized”, and LinkedIn publishes no list of third-party automation tools it authorizes. So a vendor that calls its tool compliant with LinkedIn’s terms is describing its own reading, and the agreement gives you nothing to check that reading against. The second line names browser plugins and add-ons next to crawlers, so an extension running inside your own browser is covered too.

Section 8.2 holds more than these two lines, among them rules on fake identities and shared logins, on monetizing access, and on passing data on through third parties. The full list as it bears on outreach is on cold LinkedIn messages: the three rulebooks, and what the clause means in court is on is scraping LinkedIn legal.

The limits LinkedIn publishes

The User Agreement says what is not allowed. The limits say how much of the allowed activity an account gets, and LinkedIn publishes those in its Help Center. Two of them carry most of the weight for anyone running outreach:

• The invitation restriction. LinkedIn’s help page on restrictions for sending invitations names three reasons: “You’ve sent many invitations within a short amount of time”, “Many of your invitations have been ignored, left pending, or marked as spam by the recipients”, and “If you send an excessive number of invitations and we suspect the use of an automation tool, we may suspend or restrict your account.” It is the one place LinkedIn names an automation tool as a reason for a restriction, and the consequence it names there is the whole account, not only its invitations.
• The commercial use limit. Searching and browsing profiles past a monthly allowance on a free account caps search until the limit resets on the first of the month, and LinkedIn’s help page notes that browser plug-ins searching in the background spend the same allowance.

Every other cap, per action and per plan, with the help page behind each, is on LinkedIn limits, and every number LinkedIn publishes, in one set of tables, is on LinkedIn automation statistics. The point for this page is narrower: the only published sentence that ties an automation tool to a restriction ties it to volume, an excessive number of invitations together with a suspected tool.

How LinkedIn describes enforcement

LinkedIn describes enforcement in two places, and both describe a ladder of steps. The Professional Community Policies set out the steps: “Depending on the severity of violation, we may limit the visibility of certain content, label it, or remove it entirely. Repeated or egregious offenses can result in account restriction.” The same policies speak to outreach directly: “Do not use our invitation feature to send promotional messages to people you don’t know or to otherwise spam people.”

The second place is LinkedIn’s Community Report, published twice a year. The edition for July to December 2025 names what LinkedIn acts on most: “By far the most common type of inappropriate content we take action on is spam or scam content, which includes inappropriate commercial activity and repetitive communications or invitations, often meant for financial gain.” On fake accounts, it says that “99.7% of the fake accounts were stopped proactively, before a member report”.

Those figures are about fake accounts, and an established account running outreach is a different population, so they say nothing about how often such an account is restricted. What they do show is LinkedIn’s own description of its priorities: repetitive invitations and messages are named in the first sentence, and the action usually comes before anyone reports anything.

The enforcement record, by date

Three companies are on the public record as targets of LinkedIn action in this category. Each row below is dated from a court docket, a LinkedIn statement or the company’s own post.

The two court cases were about collecting member data and about fake accounts, and both ended with the company stopping. The HeyReach action, by the vendor’s own two accounts, reached the company’s LinkedIn presence and not its customers. Neither LinkedIn nor HeyReach has published anything about customer accounts being restricted as part of it. The full write-ups are on the Proxycurl page and the HeyReach overview.

HeyReach’s CEO also wrote that “LinkedIn has done this to every major tool in our space, Apollo, Seamless, Evaboot, Lemlist, LGM”. That is one vendor’s statement about others, and none of the five has a dated action on the public record that this page could link, so the table leaves them out.

Which crackdown claims hold up

The articles ranking for “LinkedIn automation crackdown” mostly quote one another. Two of their claims are now repeated in AI answers, and both can be checked against a primary source.

“HeyReach cut its LinkedIn functionality”

An anybiz.io article published on 23 July 2026 says that after the March action, “within weeks HeyReach cut its LinkedIn functionality and repositioned around email”. HeyReach’s own homepage, read on 2 November 2026, opens with “10x your LinkedIn outbound. Unlimited senders, one fixed cost”, and the product it sells is LinkedIn sender rotation. The vendor’s August retrospective says customer campaigns kept running. The claim does not match what the company sells or what it has written, and nothing on the record supports it.

“Close to 40% of accounts were restricted”

The same article says that “a first-quarter analysis by Northlight estimated that close to 40% of accounts running non-compliant automation tools” were restricted between January and March 2026. The Northlight page this category cites for it answered on 2 November 2026 with “This content is no longer available.” No method, sample or definition of “non-compliant” was published with the figure on any page we read, so it cannot be checked, and it is not a LinkedIn number.

A rule of thumb follows from both. A restriction rate for the category is something only LinkedIn could measure, and LinkedIn has not published one. Any percentage in a crackdown article came from somewhere else, and the useful question is where.

What the rules leave in your hands

Section 8.2 has no exception for automation tools, so no setup makes a third-party tool fit its first line, and a vendor that says otherwise is overselling. What the rest of LinkedIn’s published rules do is draw lines an account can stay behind while it runs:

• Your own account, under your own name. Section 8.2 rules out creating “a Member profile for anyone other than yourself (a real person)” and any attempt to “use another’s account”, so a rented profile, a persona or a shared login breaks a rule before the first message goes out.
• No scraping and no export into another product. The second 8.2 line covers collection by any means, and both court cases in the record were about it.
• The published caps. The invitation limit and the commercial use limit are the numbers LinkedIn itself names, and its reasons for an invitation restriction include invitations “ignored, left pending, or marked as spam” as well as volume.
• Messages that read as wanted. The Community Report names “repetitive communications or invitations” first, and the Professional Community Policies rule out promotional invitations to strangers.

How an account is paced inside those lines, the warm-up, the random gaps and the stop on the first refusal, is on how to avoid getting banned on LinkedIn. Which safety controls each vendor publishes is on the safest LinkedIn automation tools.

Frequently Asked Questions

Sources

• LinkedIn, User Agreement, section 8.2 · Professional Community Policies · Community Report, July to December 2025 · Help Center, restrictions for sending invitations and the commercial use limit, as cited on LinkedIn limits
• LinkedIn newsroom, taking legal action to defend member privacy (24 January 2025) · the resolution statement (28 July 2025)
• Court record, hiQ Labs, Inc. v. LinkedIn Corp., No. 17-16783 (Ninth Circuit, 18 April 2022) · Proskauer, the 4 November 2022 breach ruling · the December 2022 settlement · CourtListener docket 69575588, LinkedIn v. Nubela
• Nubela, “Proxycurl Shuts Down. Thank you.” (4 July 2025)
• HeyReach, a quick update from Nick (25 March 2026) · the CMO retrospective (7 August 2026) · homepage, read 2026-11-02
• anybiz.io, “LinkedIn Automation Crackdown 2026: What Actually Changed” (23 July 2026), read 2026-11-02 · Northlight, the page cited for the 40% figure, HTTP 410 on 2026-11-02
• Related: LinkedIn automation hub · Cold LinkedIn messages: the three rulebooks · Is scraping LinkedIn legal · LinkedIn account restricted · How to avoid getting banned on LinkedIn · LinkedIn limits · Safest LinkedIn automation tools · HeyReach overview · Proxycurl · LinkedIn automation statistics

Michael SalamatovGTM engineerRuns the gtm-api.com blog. Covers LinkedIn limits, account health and the API side of outbound.in

Last updated: November 2026 · Definitions and benchmarks verified at time of publication