Call · 30 min
Behind the curtainMattia Esposito12 November 202611 min read

Scraping LinkedIn: is it legal? It depends on three different questions, and “the profile is public” answers none of them.

Collecting contacts from LinkedIn or any other social network is decided on three separate levels: personal data, the marketing message and the contract with the platform. Here are the rules for each, the real fines and the checklist in Word and PDF.

In brief

Public data is still personal data: you need a lawful basis to collect it, and the person must be informed at the latest with the first message. Articles 6 and 14 of the GDPR say so, and the Italian Garante repeated it for web scraping in its May 2024 note.

Writing to them to sell something has rules of its own, on social networks too. In 2021 the Garante wrote that it makes no difference whether a LinkedIn profile is open to messages: what counts is the purpose of the message.

And the contract with the platform applies separately. LinkedIn prohibits bots, crawlers and extensions for copying profiles; the CNIL fined an extension that collected contacts from it 240,000 euros.

The ten-field checklist is free to download below, in Word and PDF. This is information, not legal advice.

Written by: Mattia Esposito, AI and marketing consultant, founder of Itria AI. From Puglia, based in Turin since 2023, he works remotely in Italy and abroad. LinkedIn profile.

The checklist before collecting contacts from social media, ready to download

Ten fields to fill in before you start, each tied to the article or decision it comes from, in Word and PDF. They put in writing the decisions the GDPR expects you to be able to demonstrate: why you collect the data, on what basis, and how and when you inform the person.

The ten-field checklist to fill in before collecting contacts from social media, with the legal article behind each field
FileContentsLink
Wordto fill in, in Italian

The ten fields with room for your answers and the rule behind each.

scraping-linkedin-legale-checklist.docx

PDFto print, in Italian

The same checklist, to keep on file.

scraping-linkedin-legale-checklist.pdf

N.What to writeWhere it comes from
1

The purpose, and why it needs personal data.

GDPR art. 5(1)(c)

2

The lawful basis and, if it is legitimate interest, the balancing test.

GDPR art. 6(1)(f)

3

The source, and whether the visibility the person chose was respected.

CNIL on KASPR

4

Only the data you need, no special-category data.

GDPR arts. 5 and 9

5

What the platform’s terms say, and the decision taken.

LinkedIn, section 8.2

6

The channel for the first contact and the rule that applies.

Art. 130; PECR

7

The privacy notice, delivered at the latest with the first message.

GDPR art. 14

8

How people object, and the list of objections received.

GDPR art. 21

9

How long the data is kept.

GDPR art. 14(2)(a)

10

Who decided, when, and who checked it.

GDPR art. 5(2)

In September the founder of GenFuture, management software for the beauty sector, asked in a post whether collecting contacts from social media could be automated: he was looking for Italian creators who met precise criteria. I told him what I had learnt from trying: “The APIs of almost every social network are closed; there’s no agent that can go and fish in bulk.” But something similar was possible.

I did it in three steps: volume, then specificity, then personalisation. In 1 hour and 4 minutes, with 22 searches, 389 profiles; 30 to read one by one; 10 delivered in order of priority, in a PDF. The messages built on the signals found in their profiles got 3 replies. In his words: “Within an hour I had a PDF with all the information.” The full story is in the post of 15 September.

Work like this raises the same questions as this page: on what basis the profiles are read, what the person is told in the first message, what the platform’s terms allow. I am not passing judgement on a single case here: I am setting out the rules that apply to anyone collecting contacts, myself included.

Is scraping LinkedIn legal? The three questions

“Scraping” is not a legal category: what gets assessed are three different things, each with its own rule. A collection can be lawful under the GDPR and breach the contract with LinkedIn, or respect the contract and then lead to a message the Italian Privacy Code does not allow.

The questionWhat it asks forThe rule
Can I process this data?

A lawful basis, only the data needed, a notice to the person, the right to object.

GDPR arts. 5, 6, 14, 21

Can I write to this person to sell?

For automated promotional communications, consent; for the Garante this also covers messages on social networks.

Italian Privacy Code art. 130

Can I use these tools on this platform?

Whatever the terms of use say, and LinkedIn’s prohibit bots, crawlers and extensions.

LinkedIn User Agreement, 8.2

The three questions that decide whether collecting contacts from social media is lawful: personal data, the marketing message, the contract with the platform A public profileanswers none of them 1. Personal dataGDPR: lawful basis, minimisation, notice with the first message 2. The marketing messageArt. 130 of the Privacy Code in Italy, PECR in the UK 3. The contract with the platformTerms of use: bots, crawlers and extensions banned by LinkedIn
Three separate levels: getting one right does not answer the other two.

Public data is still personal: what the GDPR says

A name, a job title and a company on a public profile are personal data, and collecting them makes you a controller with every GDPR obligation. The Italian Garante makes the point in its 20 May 2024 note on web scraping: the obligation to identify a suitable lawful basis for processing that data remains (faithful paraphrase). The note is about generative AI, but the principle is general.

For collecting contacts the basis is almost always legitimate interest, which only holds if the person’s rights do not override it. Then minimisation: only the data needed for the purpose. Twelve authorities, the UK’s among them, put it in a joint statement in 2023: personal information made publicly accessible “is still subject to data protection and privacy laws in most jurisdictions”.

The most practical point is the notice. When the data does not come from the person, Article 14 requires telling them who is processing it, why, on what basis and from what source, “within a reasonable period after obtaining the personal data, but at the latest within one month”. If the data is used to contact them, at the latest with the first message. And if the person objects to direct marketing, the data is no longer used for it.

Writing to contacts: article 130 and messages on social media

In Italy, automated promotional communications, by email and by messages “of any other kind”, require consent, and for the Garante the rule also covers messages on social networks. Its 2013 guidelines are clear: the fact that an address is easy to find online does not authorise using it to send automated promotional communications without the recipients’ consent (faithful paraphrase).

The case closest to the question dates from 16 September 2021. An estate agency had found an owner’s name on a public register and messaged her on LinkedIn to offer a service. The Garante found the processing had no lawful basis and wrote that it made no difference whether the profile was open to messages: what counts is the purpose for which the message is sent, in this case a promotional one (faithful paraphrase).

Two clarifications, because the case is often misreported. The message itself led to a reprimand; the 5,000 euro fine was for failing to respond to the Garante’s requests. And the recipient was a private individual: for professional contacts it is an analogy, not a precedent. The rules on cold emails are covered in the guide on B2B cold email and the law.

What LinkedIn’s terms say

LinkedIn’s User Agreement prohibits developing or using software, scripts, robots, crawlers, plug-ins and browser extensions to copy the services, “including profiles and other data from the Services”. It also prohibits getting around usage limits and using information obtained from LinkedIn through third parties, such as aggregators and brokers, without consent.

The help page on prohibited software points the same way: no third-party tools that collect data, change how the site looks or automate activity. The contract does not depend on the GDPR: it applies even when processing the data would be lawful in itself. And in the United States, this is the point on which LinkedIn prevailed against hiQ.

The real fines, and what caused them

The fine closest to scraping LinkedIn is the one the French CNIL imposed on KASPR: 240,000 euros, on 5 December 2024. KASPR sold a Chrome extension that showed the contact details of the people you visited on LinkedIn and built a database, around 160 million contacts according to the CNIL.

CaseWhat was foundOutcome
KASPR (CNIL, France, 2024)

Contacts taken even where the user had limited their visibility; data kept for too long; a privacy notice only from 2022, four years after launch.

240,000 euros and deletion

La Prima (Garante, 2021)

A sales message on LinkedIn to a private individual found on a public register, with no lawful basis.

Reprimand; 5,000 euros for failing to respond to the Garante

Clearview AI (Garante, 2022)

Images collected from the web, social networks included, turned into biometric data.

20 million euros

hiQ v LinkedIn (United States, 2022)

The court of appeals upheld a preliminary injunction on access to public profiles; on the contract, LinkedIn later prevailed.

Settled with a consent judgment

The KASPR case shows where the practical line lies. The CNIL did not fine anyone for looking at a public profile: it fined the collection of contact details of users “qui avaient pourtant choisi d’en limiter la visibilité” (who had chosen to limit their visibility), keeping them too long and not saying so. Field 3 of the checklist comes from there. Clearview is a different case, because biometric data has stricter rules: it is only useful as an outer limit.

Italy, the European Union, the United Kingdom and the United States

In Italy, article 130 requires consent even to write to a business address, while in the UK you can write to companies without consent, but not to sole traders. In the United States the question has been fought more over access to websites and the contract than over privacy.

WherePublic data and the marketing messageSource
Italy

Public data is still personal; an automated promotional message, on social networks too, requires consent, and for the Garante this also applies to legal persons.

GDPR; art. 130; Garante 2013 and 2021

European Union

Lawful basis, minimisation, notice with the first message, objections always honoured.

GDPR arts. 5, 6, 14, 21

United Kingdom

Public data is still protected; by email or social media message you can write to companies, while sole traders are treated as individuals.

UK GDPR; PECR; ICO

United States

No general federal privacy law read for this page; in the hiQ case the crux was access and the contract.

hiQ v LinkedIn, 2022

The ICO says so for direct messages: the rules cover emails and “direct messages via social media”, and “Sole traders and some partnerships are treated as individuals”. For Italy there is no recent decision specifically on professional contacts between businesses: it is an open point, and the checklist treats it as one.

Scraping LinkedIn with n8n or an extension

The tool changes the method, not the three questions: an n8n flow, a script or a browser extension still falls under the ban in LinkedIn’s terms, and the data they collect is still personal data. The CNIL fined precisely a Chrome extension.

An automation works well where the data belongs to the business or comes with people’s permission: incoming enquiries, contacts who have already written, the CRM.

How to find clients on social media within the rules

The rules read for this page all point the same way: less data, better chosen, and a first message that says who you are, why you are writing and how to stop the messages. It is also what works: in the case described above, 389 profiles became 10 people, and the messages built on their signals got 3 replies.

The three-step method, volume, specificity and personalisation, is explained in the guide to B2B lead generation, along with what you can do with the data. The lead entry in the glossary explains the terms. The other tests with numbers are in the Behind the curtain section, including 521 hours in the markets with AI agents and the theme page with 632,288 likes. The full sequence, from the source to the first message approved by a person, is in the guide on using AI to find customers.

Questions and answers

Is scraping LinkedIn legal?

It depends on three separate questions: whether there is a lawful basis for processing the personal data (GDPR), whether the marketing message that follows complies with article 130 of the Italian Privacy Code, and whether you respect LinkedIn’s terms, which prohibit bots, crawlers and extensions for copying profiles. A public profile answers none of the three. This is information, not legal advice.

If data is public on LinkedIn, can I use it?

It is still personal data. For the Italian Garante, the fact that an address is easy to find online does not authorise its use for promotional messages without consent, and in the 2021 La Prima case it wrote that it makes no difference whether the profile is open to messages: what counts is the purpose of the message.

Can I message a LinkedIn contact to sell them a service?

For the Italian Garante, promotional messages on social networks, private or on a feed, fall under the rules on unsolicited communications. In the La Prima case an agency messaged a private individual on LinkedIn after finding her on a public register, and the Garante found there was no lawful basis. She was a private individual: for professional contacts it is an analogy, not a precedent.

Does using n8n or a browser extension change anything?

It changes the method, not the questions. LinkedIn’s terms prohibit software, scripts, robots, crawlers, plug-ins and extensions for copying profiles, and its help page bans third-party software that automates activity. The CNIL fined an extension that collected contacts from LinkedIn 240,000 euros.

How soon must I inform someone whose data I have collected?

Article 14 of the GDPR sets a reasonable period, at the latest one month; if the data is used to contact the person, at the latest at the first communication. The notice also states where the data came from and how to object, and an objection to direct marketing must always be honoured.

Notes on sources

  1. Regulation (EU) 2016/679, GDPR, Articles 5, 6, 14 and 21: the quotation on the privacy notice comes from here. Read on 12 November 2026.
  2. Legislative Decree 196/2003, art. 130, updated to 9 November 2026 (in Italian).
  3. Garante, injunction order against La Prima S.r.l., 16 September 2021; guidelines on spam, 4 July 2013; note on web scraping, 20 May 2024; Clearview AI, 10 February 2022 (all in Italian; quoted in faithful paraphrase).
  4. CNIL, fine on KASPR, decision of 5 December 2024; the EDPB news item, 23 January 2025.
  5. LinkedIn, User Agreement, section 8.2, effective 3 November 2025, and prohibited software and extensions.
  6. ICO, electronic mail marketing; joint statement on data scraping, 24 August 2023, signed by twelve authorities, including the ICO but not the Italian Garante.
  7. U.S. Court of Appeals for the Ninth Circuit, hiQ Labs v. LinkedIn, 18 April 2022. The December 2022 consent judgment is reported by secondary sources and was not read in the original.
  8. The 389-profile case comes from the post of 15 September 2026 and from the documented cases on the home page, published with the client’s permission.
·Behind the curtain

Finding the right people is the easy part. Writing each of them something that holds up, within the rules, is the work.

This page is part of Behind the curtain: the things I have actually done, with the numbers, and the rules around them. Mattia Esposito, founder of Itria AI.