LinkedIn BrowserGate: The Secret Extension Scanner Exposed

doc

BrowserGate scandal: LinkedIn injects a 2.7MB JavaScript bundle called 'Spectroscopy' that probes 6,000+ Chrome extensions, collects 48 device characteristics, and encrypts the fingerprint for every session — entirely absent from their privacy policy. Here's what it means for your privacy.

LinkedIn was disclosed on April 5, 2026 by Fairlinked e.V., with the practice tied to Teamfluence Signal Systems OÜ, as running an extension-scanning fingerprinting system inside pages served to Chromium-browser visitors. The site delivers a roughly 2.7 MB JavaScript bundle, internally referenced as APFC or DNA, that probes installed browser extensions, reads about 48 device characteristics, encrypts the resulting fingerprint, and attaches it to every API request in the session. The practice was not described in the company's privacy policy, and it affects every Chromium-based browser rather than Chrome alone.

What the bundle does

Anti-fraud systems on large sites commonly check for signs of automation, scrapers, and altered clients. The LinkedIn implementation goes further than behavioral checks: it builds a model of the visitor's browser configuration and device, then carries that model through the whole session. The fingerprint is injected into the HTTP headers of each API request, so every subsequent call to the platform is associated with the same extension and device profile.

The collected profile combines two kinds of signal. Extension detection identifies which installed add-ons the visitor has active; device characteristics describe the machine and browser environment itself. Together they produce a far more stable identifier than a cookie, because extensions and hardware details change only when the user changes them.

The three detection stages

  1. Extension resource probing — the bundle checks for chrome-extension:// resources associated with known extension IDs, inferring which add-ons are installed by whether those resources respond.
  2. DOM element inspection — a lighter check looks for extension-injected elements and page modifications that reveal active add-ons without a full tree walk.
  3. Full DOM tree scan, called Spectroscopy — the complete page document is walked to detect extension footprint that the lighter stages would miss.

The scanned list covered 6,222 extensions in the disclosed analysis. An independent retest by BleepingComputer observed 6,236 extensions, a small difference consistent with list updates between measurements.

What was collected and where it went

Alongside the extension list, the bundle gathered about 48 device characteristics. These include CPU core count, memory amount, screen resolution, multiple-monitor setup, timezone, language, battery status, audio characteristics, and storage details.

The assembled data was encrypted with RSA before transmission. It was sent to LinkedIn servers and to HUMAN Security, formerly PerimeterX, the third-party fraud vendor involved in the system. Encryption protects the data in transit; it does not change what the recipient is permitted to do with the decoded fingerprint.

How the extension list grew over time

Point in timeExtensions on the scanned list
201738
2024461
December 20255,459
February 20266,167

The growth shows the list moving from a small set of known scraping tools to a broad catalog covering much of the installed extension ecosystem. Each addition widens the inference the fingerprint can make about the visitor.

Why the extension list is more sensitive than a cookie

A cookie is an identifier assigned by the site; the extension list is a readout of personal software choices. Extensions installed for faith or community, political orientation, accessibility needs, disability-related tools, job hunting, and competitor sales activity each map onto protected or sensitive personal attributes. The disclosed list covers categories in those areas, along with sales and competitor tools.

Unlike a cookie, which can be cleared, the extension profile reflects what a person actually uses day to day. Inferring religious affiliation, political leaning, or accessibility needs from installed add-ons turns browsing configuration into a proxy for sensitive characteristics, and one that the visitor typically does not realize is being enumerated.

Where the legal and privacy debate sits

The controversy runs on two tracks. Under privacy rules such as GDPR and the ePrivacy framework, collecting device and software details without clear disclosure raises questions about lawful basis and notice; the fact that the practice was absent from the privacy policy is central to that argument. Separately, the Digital Markets Act frames how large platforms observe and restrict users in their own services, which adds a regulatory layer beyond conventional privacy law.

The legitimate side of the practice is real: large platforms do face scrapers, fake accounts, and automation abuse, and fraud vendors need signals to tell those apart from normal visitors. The disputed line is between checking for automation and silently building an extension-and-device dossier that reveals sensitive attributes.

What readers can do

  • Switching to another Chromium browser does not remove the exposure; the detection targets the shared browser engine, so visitors on any Chromium-based client are in scope.
  • Reduce extension exposure by removing add-ons no longer used; fewer installed extensions means fewer signals in any profile built this way.
  • Check what the privacy policy and the third-party fraud vendor disclose about device and extension collection before connecting business or personal accounts.
  • Enterprise networks should review whether employee sessions to major platforms are accumulating extension-level fingerprints tied to work identities, and include such collection in vendor privacy reviews.

Frequently asked questions

Does this mean LinkedIn can see my exact extension names?

The system probes a catalog of known extensions and records which ones appear present, along with device characteristics. It builds a match against that list rather than reading arbitrary extension internals; the matching result still reveals which listed add-ons the visitor has.

Does using a non-Chromium browser protect me?

The disclosed practice targeted Chromium-engine visitors. Clients on other engines were not part of this particular system, though they may still be subject to ordinary fingerprinting and fraud checks; the engine switch changes the signal available, not the site's general anti-fraud behavior.

Why is extension collection treated more seriously than ordinary device fingerprinting?

Because the extension catalog can proxy sensitive attributes such as faith, political views, disability, or job search status, rather than just describing hardware. A resolution or timezone signal identifies a machine; the right extension list identifies personal circumstances.

LinkedIn BrowserGate: The Secret Extension Scanner Exposed