The advertising infrastructure that follows people across websites and apps has now been publicly linked to something far more invasive than personalized marketing: silently delivering spyware to selected devices.
Researchers at the University of Toronto’s Citizen Lab say they have documented the first publicly confirmed case of zero-click mercenary spyware being delivered through targeted online advertisements. The technique, which researchers call “AdFection,” can infect a target after an advertisement is displayed without requiring the person to click a link or interact with the ad.
Citizen Lab’s investigation centers on Discover-Tech, an Israel-based company researchers describe as connecting the commercial advertising ecosystem with mercenary spyware vendors. Independent reporting by Reuters found that Discover-Tech’s advertisements appeared across more than 120,000 websites and about 75,000 apps.
Real-time advertising can identify a target before an attack
Modern digital advertising operates through automated auctions.
When someone opens a website or app, information about the device and user can be broadcast to companies competing to display an advertisement. Citizen Lab says this ecosystem processes hundreds of billions of bids and ad sales each day, exposing information that may reveal browsing behavior, movements, interests and device characteristics.
The researchers found that this infrastructure could be used not simply to decide which advertisement someone sees, but to identify a specific device and send malicious code directly to it.
Citizen Lab said HTML and JavaScript delivered with an advertisement can retrieve additional code, including browser exploits capable of installing spyware. Because advertising systems contain identifiers tied to devices or users, malicious content can potentially be delivered to a highly specific target.
That turns technology originally built for precision marketing into infrastructure for precision surveillance.
At least 45 Al Jazeera devices were infected
The investigation linked the technique to spyware developed by Candiru, an Israeli surveillance company that was added to a U.S. government trade blacklist in 2021.
Citizen Lab said it confirmed with high confidence that 45 devices belonging to the international media network Al Jazeera were infected using Candiru spyware during 2024 and 2025.
Reuters also reported that advertising records showed Discover-Tech-linked activity reaching networks associated with sensitive organizations, including NATO, the Pentagon and the United Nations. The evidence that advertisements reached those networks does not by itself establish that every device was compromised.
Security researchers from Confiant, which worked with Citizen Lab, also identified activity involving networks associated with organizations including the International Atomic Energy Agency and International Criminal Court.
The weakest link may be the advertising supply chain
The findings expose a cybersecurity problem larger than one spyware vendor.
People generally think of zero-click attacks as vulnerabilities delivered through messaging applications, operating systems or specially crafted network traffic. AdFection demonstrates that the advertising stack itself can become a delivery mechanism.
Google and other advertising companies have responded by removing or suspending Discover-Tech from parts of their ecosystems, while IAB Europe suspended the company from its approved vendor list following the findings.
For developers and security teams, the case changes the threat model around third-party advertising code.
An advertisement is not merely visual content. It can involve scripts, identifiers, auction infrastructure and external vendors executing within applications and browsers.
That makes ad tech part of the security perimeter.
The uncomfortable lesson is that the infrastructure designed to identify exactly which person should see an advertisement can also help an attacker determine exactly which person should receive an exploit.