Socket links 319 browser extensions to search hijacking affecting 13.4 million users over a decade

Socket links 319 browser extensions to search hijacking affecting 13.4 million users over a decade

Browser extension operators diverted searches through their own servers using 319 extensions, according to an investigation by Socket. The extensions reached roughly 13.4 million users over a decade, with records dating back to 2016.

Socket’s October 29 report, written by John Tuckner, added 264 extensions affecting 8.7 million users to earlier research. Many were promoted as tools for finding manuals, converting PDFs, browsing safely or searching with AI.

The redirects expose search queries to the operators before users reach their results. Socket said those queries are often associated with unique identifiers, allowing operators to build detailed search histories.

Operators also control where each request goes. That creates a possible route to phishing pages or malicious downloads, although Socket did not establish that such attacks occurred.

Search settings turn queries into referral income

The extensions exploit a standard browser feature rather than requiring complex code. An extension’s manifest file can replace the default search provider through the chrome_settings_overrides setting.

Socket highlighted Safe Game Search as an example. Its entire package consisted of a manifest and three icon files, without requested permissions, background logic, content scripts or an interface.

Installing an extension that changes the search provider triggers a prompt. However, Socket said that prompt does not clearly explain where searches will ultimately travel.

The configured provider sends queries to an operator-controlled server, which then redirects them to a search engine. Yahoo and Microsoft share advertising revenue with partners that supply search traffic.

Tracking parameters identify the partner receiving credit. Yahoo uses hspart= and hsimp=, while Bing uses PC= and PTAG=.

Socket observed mostly redirects that added these partner tags before delivering search results. But the intermediary servers could change destinations without any change to the installed extension.

Predator Search, which had 50,000 users, illustrated that flexibility. Six identical searches through its proxy produced four different outcomes.

One went directly to Yahoo with partner tags. Another reached searchlio.net, while a third passed through searchtosearch.com and mcftink.com before reaching Yahoo.

The fourth outcome was a Cloudflare error page. Socket also observed other cases where repeated searches reached Google’s homepage while different queries continued to Bing.

These findings demonstrate control over individual redirects, not confirmed delivery of malicious pages. Socket identified query collection and referral accounting as possible reasons for using intermediary servers.

Advertising brings users to unlisted extensions

Most extensions in the investigation were unlisted in the Chrome Web Store, Socket said. Their pages remained accessible through direct links, but store searches and other indexers did not expose them.

Paid advertisements supplied visitors instead. Socket traced promotions through Google Ads, Bidbrain and other networks, often using popovers or popunders on webpages.

Those ads led to branded installation pages, sometimes followed by automatic redirects to the Chrome Web Store. Brands such as Safe Web Search and PDF Manual Finder presented the extensions as narrow-purpose utilities.

Socket linked 200 extensions, accounting for about 5.68 million users, to identified operators. Another 119 extensions, with roughly 7.76 million users, remained associated with unknown operators.

The largest identified group was Intango LTD, with 111 extensions and approximately 2,187,000 users. Socket mapped its activity across about 279 domains, often using separate domains and administrative email addresses for individual brands.

Among unattributed groups, SecuryBrowse accounted for 58 extensions and approximately 4,901,000 users. Those extensions often sought sensitive permissions beyond changing search settings.

A newer group, labelled Bing-Search, comprised 37 extensions with approximately 2,210,000 users. Socket said shared Bing partner tags and closely timed updates indicated a single operator behind apparently separate publishers.

Fifteen publishers in that group issued updates within the same four days in June 2026.

Tracing the operators

Socket combined extension listings, privacy policies and Google Ads Transparency records to connect brands with advertising buyers. It also used Censys, Urlscan and DecryptAds to map installation pages, redirects and advertising relationships.

Those methods have limits. DecryptAds maps advertising sellers but cannot identify the advertisers purchasing installation ads, while some privacy policies omit company names.

Socket found that domains sometimes remained active after their extensions disappeared from the store. The report also documented new domains and installation lures appearing after removals, extending a pattern first observed in 2016.

Got Something To Say?

Your email address will not be published.