All In
FR EN
Back to blog
IA & outils pour LinkedIn 18 min read

LinkedIn AI Prospecting Tools: What US Privacy Law Requires

| By Patrick de Carvalho

Contents

I get the same message every few weeks: which AI tool should the sales team use to prospect on LinkedIn without getting the company sued or banned. Wrong question, asked in the wrong order. In the United States there is no single law to check a tool against. There is a federal void, nineteen state laws with different definitions of your data, an FTC that polices unfairness rather than scraping specifically, and a LinkedIn contract that says no to automation in plain English regardless of which of the above applies to you. I read the statutes, the court ruling every law firm blog cites and half get wrong, and the pricing pages of five tools sales teams run today. Here is what holds up.

In short: The United States has no federal privacy law equivalent to GDPR. California's CCPA/CPRA is the outlier reaching B2B contact data (the employment/B2B exemption expired December 31, 2022), and its Delete Act now requires anyone who collects and sells personal information, enriched LinkedIn data included, to register as a data broker starting in 2026. hiQ Labs v. LinkedIn (2019-2022) established that scraping public LinkedIn data does not violate the federal Computer Fraud and Abuse Act, but LinkedIn still won on a separate ground: breach of its own User Agreement. That is the rule that actually gets accounts banned.

What counts as an AI LinkedIn prospecting tool in 2026

Three products get sold under the same "AI prospecting" label, with three risk profiles: a browser extension running inside your live LinkedIn session in Chrome, clicking and sending while the tab is open; a cloud platform running campaigns from a remote server, often through a dedicated IP, no browser required; and a data platform that enriches contact records and drafts outreach copy with a language model but never touches LinkedIn's interface directly.

"AI" here means two things: a language model wrapper drafting connection notes and follow-ups, and a scoring layer ranking which enriched contact is worth a human's time. The generic version of the first feature is why a large share of long-form LinkedIn posts get flagged as fully machine-written; Pangram Labs put that figure above 40% in a July 2026 sample, covered in the LinkedIn AI slop button piece. This article is about what happens before a message gets sent: how the contact was obtained, under what legal authority.

There is no US GDPR, and the gap is not empty space

The fact that surprises non-US teams most: the United States has no comprehensive federal privacy statute. DLA Piper's tracker of global data protection law states it without qualification: no national privacy law in the country, only sector rules for health records, financial data, telecommunications, and children's data.

What exists instead is a state-by-state patchwork. By DLA Piper's mid-2026 count, nineteen states have comprehensive consumer privacy laws in force, starting with California and now including Virginia, Colorado, Connecticut, Utah, Texas, Oregon, and a dozen others that followed between 2023 and 2026. Each defines "personal information," "sale," and "consumer" slightly differently, with its own compliance thresholds.

The detail that changes the calculus for LinkedIn prospecting: most of these nineteen laws exclude employment and B2B data, because their definition of "consumer" is a natural person acting in a personal or household capacity, not someone's work title and company email. California is the exception. A sales team scraping work profiles for outreach into Texas or Virginia may sit entirely outside consumer privacy law today; the same activity aimed at a California-based contact does not.

The California Consumer Privacy Act became law in 2018 and was expanded by Proposition 24, the California Privacy Rights Act (CPRA), in November 2020. Its stronger provisions took effect January 1, 2023, enforced by the California Privacy Protection Agency (CPPA).

Two changes matter here. First, the exemptions carving out employment records and B2B communications expired December 31, 2022, per the California Attorney General's office. Since then, a prospect's work email and job title, collected for a sales sequence, count as personal information under California law like consumer data, with the same notice and opt-out obligations attached.

Second, and newer: the Delete Act. The CPPA defines a data broker as "a business that collects personal information about consumers from various sources and sells that information to other companies." Any organization fitting that description must register annually, during a January window, disclosing what data categories it collects, who it sells to, and how many consumer requests it received the prior year. As of January 1, 2026, California residents can file a single deletion request through the state's Delete Request and Opt-out Platform, DROP, reaching every registered broker at once; brokers must start processing those requests by August 1, 2026.

Weigh the enrichment layer of the prospecting stack against that definition: a platform that waterfalls contact lookups through more than a hundred data providers and resells the combined record to its customers is describing something close to what the Delete Act calls a data broker. Whether that duty falls on the vendor, the company buying the data, or both is an open question, flagged again in "What I don't know" below.

What the FTC actually polices, and what it does not

I looked for a Federal Trade Commission rule aimed specifically at scraping or automated LinkedIn prospecting. I did not find one.

What the FTC does have is Section 5 of the FTC Act, a general-purpose statute barring "unfair or deceptive acts or practices" across nearly every industry. It is the tool the agency reaches for when a company misrepresents how it collected data, ignores a privacy policy it published, or makes false claims about what its AI product does. None of that is specific to LinkedIn, scraping, or sales automation; it is the same standard that applies to a mattress ad. The limits that apply come from contract law and the states that chose to legislate, not a scraping-specific federal rule.

hiQ Labs v. LinkedIn: the ruling everyone misquotes

This is the case every "is scraping legal" article cites, and most stop reading a paragraph too early.

hiQ Labs built analytics products from public LinkedIn profile data. LinkedIn sent a cease-and-desist and blocked hiQ's access; hiQ sued for an injunction, arguing LinkedIn was misusing the Computer Fraud and Abuse Act (CFAA) to shut down a competitor. In 2019, the Ninth Circuit sided with hiQ: scraping data that is publicly accessible, with no login wall, does not violate the CFAA, because there was no unauthorized access to begin with.

The Supreme Court vacated that ruling on June 14, 2021, and remanded, citing Van Buren v. United States, which narrowed the CFAA's "exceeds authorized access" language to someone with legitimate system access reaching into an off-limits area, not someone who simply violates a website's posted terms. In April 2022, the Ninth Circuit reconsidered under that narrower standard and reached the same conclusion: scraping public data still did not violate the CFAA.

The part most write-ups leave out: on remand, the district court in the Northern District of California ruled that hiQ had separately breached LinkedIn's User Agreement, a contract claim entirely independent of the CFAA fight. The parties settled in November 2022. hiQ won on federal hacking law and lost on contract law, and the second result is the one that shut the company down.

LinkedIn's own contract is the rule that bites first

LinkedIn's User Agreement, section 8.2, spells out what the platform will not tolerate. Four clauses matter here. It bars anyone from developing or using "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." It separately bars overriding security features or use limits. It bars unauthorized redistribution of data obtained through the platform, including through "third parties (such as search tools or data aggregators or brokers)." And it names automation by function, not by tool brand:

"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."

None of that depends on state privacy law, FTC rules, or the CFAA. It is a contract every LinkedIn account holder agrees to at sign-up, and LinkedIn enforces it unilaterally: restriction, suspension, and litigation are available without waiting for a regulator. That is the ground it stood on against hiQ once the federal hacking claim collapsed.

Tool vendors know this, and say so in writing. Expandi's terms state that it "is not an official product published by LinkedIn" and that use of the tool "is at your own risk." HeyReach's footer carries the same disclaimer: the company "is not associated with, or endorsed by, the LinkedIn Corporation." That is the vendor telling you, before you pay, that the contract risk sits with you.

Five tools sales teams are actually running, compared

Five products cover most of what a US B2B sales team runs in 2026.

Tool Architecture Entry price What it actually does
Apollo.io Cloud data platform Free tier, paid plans scale by credits Sales intelligence database, 600,000+ companies; contact search, enrichment, sequencing; terms forbid reselling Apollo data
Clay Cloud enrichment + AI agents Free (500 actions/month), $167/month Launch plan Waterfall enrichment across 150+ data providers, "Claygents" AI research agents, syncs to LinkedIn and CRM tools; SOC 2 Type II and ISO 27001/42001 certified
Expandi Cloud automation, dedicated IP $99/month ($79 annualized) LinkedIn outreach automation from a remote server, country-based dedicated IP, automated warm-up
HeyReach Cloud automation, multi-account Paid, pricing on request Rotates outreach across unlimited connected LinkedIn accounts; integrates with Clay, HubSpot, n8n; publishes a Data Processing Addendum
Dux-Soup Browser extension, cloud option $14.99/month (Pro Dux), $99/month (Cloud Dux) Connection and messaging automation in-browser at entry tier, managed cloud option at top tier

A position: prefer the cloud architecture over a bare browser extension for anything beyond a handful of daily actions. An extension executes inside your own logged-in session, close to the "scripts... browser plugins and add-ons" language in LinkedIn's prohibition. A cloud tool at least moves execution off your browser's native fingerprint. Neither makes the activity compliant with LinkedIn's contract; one is harder to attribute to you by accident.

The daily numbers that decide whether your account survives

No vendor sets these limits. LinkedIn does, and every automation tool operates underneath them whether it says so or not. Evaboot's 2026 tracking puts LinkedIn's weekly connection request allowance at roughly 100 for a standard account, rising toward 200 for a high Social Selling Index and strong acceptance rate: about 20 a day, with a hard ceiling of 30,000 total first-degree connections.

Action Free / Standard Premium Sales Navigator
Connection requests ~100/week ~100 to 200/week (SSI-dependent) ~100 to 200/week (SSI-dependent)
Direct messages Up to 150/day Up to 150/day Up to 150/day
InMail Not available Up to 25/day Higher, plan-dependent
Profile visits 80/day 150/day Up to 1,000/day

A tool advertising 300 or more connection requests a week, a number pulled from Expandi's own marketing copy for one campaign type, describes volume at or above LinkedIn's published ceiling. Read that as a warning label before a feature.

The UK and the EU, briefly, because some of you sell there too

The doctrine inverts once a prospecting list crosses the Atlantic. In the United Kingdom, no CFAA-style computer-misuse claim does the heavy lifting. The relevant framework is UK GDPR, the Data Protection Act 2018, and the Privacy and Electronic Communications Regulations for unsolicited marketing messages, enforced by the Information Commissioner's Office. Most UK-directed B2B outreach relies on the same basis available across the EU: legitimate interest, under GDPR Article 6(1)(f), accepted by most European data protection authorities, France's CNIL among them, for B2B prospecting provided the prospect can object and receives basic processing information. Penalties reach €20 million or 4% of global annual turnover, whichever is higher, well above anything the California Delete Act currently threatens.

The structural difference for a US-based team: EU and UK law regulate the content of the outreach message itself, not only how the contact data was obtained. Nothing in the US framework above does that.

A stack you can run Monday morning, free tier included

None of this requires a purchase order before Monday. The free path is LinkedIn's native limits, respected manually, no extension, no cloud tool: roughly 20 connection requests a day, sent by a person, from the standard interface. It costs nothing, breaks no clause in section 8.2, and is where every configuration below should start.

Once volume outgrows what one person can send by hand, Apollo's free tier covers verified email lookups without a scraping tool touching LinkedIn directly, and Clay's free tier (500 actions, 100 data credits a month) covers light enrichment. Move to a cloud automation tool, Expandi or HeyReach, only once daily volume exceeds manual capacity, not on day one because a demo looked good. Before paying, read the vendor's disclaimer page: "use at your own risk" and "not affiliated with LinkedIn" are the vendor's risk assessment, priced in and handed to you in writing.

One more check before scaling: if the activity grows into reselling enriched contact data, ask counsel whether the Delete Act's data broker registration duty reaches your business, not just the vendor's. That question has no marketing page written for it yet.

What I don't know

I did not find a Federal Trade Commission enforcement action brought specifically against a LinkedIn prospecting tool or its users; the agency's general Section 5 authority exists, a targeted case tied to this exact activity does not appear to, at least not one I could verify.

I don't know how the nineteen state privacy laws, individually, would treat LinkedIn profile data as "publicly available information," a category most of them exempt from full consumer rights. States differ on whether that exemption covers anything beyond government records, and LinkedIn profile data is not a government record. I would not assert an answer either way without reading each statute's exemption clause directly, which is outside what I can responsibly do here.

I don't have a public figure for how often LinkedIn actually enforces section 8.2 against automation tool users, what share of flagged accounts get restricted versus terminated, or how that rate has moved since the "Seems like AI slop" reporting button shipped in July 2026. If a reliable number surfaces, it belongs in the All In newsletter, not asserted here without it.

And I don't know, with confidence, whether a mid-size sales team buying enrichment through Clay or Apollo, rather than building its own database, would itself trigger Delete Act data broker registration if it later resells or shares that enriched data downstream. That reads as an open legal question rather than a settled one, and I would treat any confident answer to it, including my own instinct, with suspicion until a lawyer who tracks CPPA guidance says otherwise.

The compliance question is really a contract question

The finding that surprised me: the compliance panic in most sales team Slack channels centers on privacy law, and the fastest, most certain risk is not privacy law at all.

A state attorney general or the CPPA might act on a CCPA violation months or years after the fact, through an investigation. LinkedIn does not need any of that. Section 8.2 lets it restrict or terminate an account tomorrow, the same clause that ended hiQ Labs' automated business after the CFAA claim had gone in hiQ's favor. Fix the contract exposure first. The privacy law question is real, California in particular, but it is the second problem, not the first.

FAQ

Scraping publicly accessible LinkedIn data does not, by itself, violate the federal Computer Fraud and Abuse Act; hiQ Labs v. LinkedIn (Ninth Circuit, 2019, reaffirmed 2022 after the Supreme Court's Van Buren ruling) established that. The same case ended with hiQ found to have breached LinkedIn's User Agreement, a separate and, in practice, faster-moving legal exposure.

Does the CCPA apply to LinkedIn contact data used for B2B sales prospecting?

Yes, for activity connected to California. The CCPA/CPRA's exemptions for employment and business-to-business personal information expired December 31, 2022, per the California Attorney General's office. Prospect and company contact data now counts as personal information under the law, carrying the same notice and opt-out obligations as consumer data.

Can LinkedIn suspend my account for using an automation tool even without breaking any privacy law?

Yes. LinkedIn's User Agreement, section 8.2, separately prohibits scraping, bots, and automated engagement, independent of any state or federal privacy statute. That contract clause is what LinkedIn enforces day to day, and it is the ground it won on against hiQ Labs after the CFAA claim failed.

What is California's Delete Act, and does it apply to LinkedIn data enrichment tools?

The Delete Act requires businesses that collect personal information and sell it to other companies to register annually as data brokers with the California Privacy Protection Agency. Starting January 1, 2026, California residents can request deletion from every registered broker through the DROP platform. Whether a specific enrichment vendor qualifies depends on its own resale practices.

Are AI LinkedIn prospecting tools GDPR compliant for outreach into the EU?

Compliance depends on the legal basis the sending company uses, not on the tool. Most B2B outbound into the EU relies on legitimate interest under GDPR Article 6(1)(f), which requires that the prospect can object and receives basic information about the processing. No vendor is GDPR compliant by default.

What is the difference between a browser extension and a cloud-based LinkedIn automation tool?

A browser extension runs inside your own LinkedIn session in Chrome, the exact pattern LinkedIn's User Agreement names when it prohibits scripts, plugins, and add-ons. A cloud tool runs campaigns from a remote server, often through a dedicated IP address, independent of whether your browser is open.

Does the FTC regulate LinkedIn scraping or prospecting automation directly?

Not with a rule specific to this activity. The FTC's authority comes from Section 5 of the FTC Act, which bars unfair or deceptive practices generally, with no dedicated scraping or LinkedIn-automation provision. The practical limits on this activity come from LinkedIn's contract and state privacy statutes, not federal trade regulation.

Sources

  1. LinkedIn User Agreement, section 8.2: scraping, bot, and automated-engagement prohibitions, quoted directly.
  2. hiQ Labs, Inc. v. LinkedIn Corp., case history, Wikipedia: 2019 Ninth Circuit ruling, the Supreme Court's 2021 remand citing Van Buren v. United States, the April 2022 reaffirmation, and the November 2022 settlement following the breach-of-contract finding.
  3. California Department of Justice, Office of the Attorney General, California Consumer Privacy Act (CCPA), oag.ca.gov, 2026: expiration of the employment and business-to-business exemptions on December 31, 2022, and current business obligations.
  4. California Privacy Protection Agency, Data Broker Registry, cppa.ca.gov, 2026: data broker definition, annual registration window, and the DROP deletion platform, effective January 1, 2026, with processing required by August 1, 2026.
  5. DLA Piper, Data Protection Laws of the World, United States, dlapiperdataprotection.com, 2026: absence of a federal privacy law, count of nineteen state comprehensive privacy laws, and California's outlier status on B2B and employment data.
  6. Apollo.io, Pricing: plan structure, scale (600,000+ companies served), and data resale restrictions.
  7. Clay, Pricing: plan structure, waterfall enrichment across 150+ providers, and SOC 2/ISO/GDPR/CCPA compliance claims.
  8. Expandi, Pricing: plan structure, dedicated country IP, automated warm-up, and its own liability disclaimer.
  9. HeyReach: multi-account architecture, integrations, and its own LinkedIn-affiliation disclaimer.
  10. Dux-Soup, Pricing: extension and cloud tier structure and pricing.
  11. Evaboot, LinkedIn Limits for Connection Requests & Messages (2026): current connection, messaging, InMail, and profile-visit thresholds by account tier.
  12. Pangram Labs, AI in your feed report, July 2026: share of long-form LinkedIn posts flagged as fully AI-generated.

All In: the contract you signed matters more than the law you didn't break

Privacy statutes decide what a regulator can do to your company months from now. LinkedIn's own User Agreement decides what happens to your account tomorrow, and it is the document almost nobody reads before installing a prospecting tool.

All In is the B2B media that decodes LinkedIn, expert blog, weekly podcast and newsletter for SME leaders and sales directors who want to turn LinkedIn into measurable growth. An original creation by Patrick de Carvalho, on LinkedIn since 2004. Motto: "I Never Lose."

Discover All In and get what matters on LinkedIn, every week.