45 Devices Hacked by Just Loading an Ad [2026]

Citizen Lab, the internet watchdog housed at the University of Toronto, published a 72-page report on October 15, 2026 that documents a spyware delivery method researchers had never confirmed in public before: hacking a phone or laptop through an ordinary online advertisement, with no click, no download and no visible sign anything happened. The lab calls the technique AdFection, and its report traces the infrastructure back to an Israeli advertising company called Discover-Tech and spyware built by Candiru.

The finding landed as one of the most significant cybersecurity stories of the year, confirmed across outlets including TechCrunch, which first reported that governments are abusing online ad networks to hack targets at scale. Citizen Lab itself described the research as the first public technical confirmation of a zero-click mercenary spyware attack delivered via targeted advertising, a milestone that moves the surveillance industry’s most dangerous capability out of courtroom filings and into the open.

Google · Preferred Sources

Don't miss new tech stories on Google

Add TrendinTech once in the Google app and our stories appear in your news suggestions.

Add Now

What Citizen Lab Found in Its 72-Page Report

The report, published directly on Citizen Lab’s research page, lays out how an advertising platform built for marketers was repurposed into a targeting and infection pipeline for government clients. Citizen Lab says it is still unclear exactly which governments commissioned the attacks it documented, a detail the lab was careful to flag as unresolved rather than speculate on. What the report does establish with specificity is the chain of companies and code involved: an ad-tech firm that profiled and served the ads, and two separate spyware vendors that turned those ad impressions into device compromises.

Researchers reviewed four billion advertising records as part of the investigation, a dataset large enough to map how targeting worked across markets rather than relying on a handful of recovered phones. That volume of records let Citizen Lab trace a pattern of malicious ad delivery that it says ran across nearly 200,000 websites and mobile apps over four months, reaching audiences in dozens of countries. The scale is part of what sets AdFection apart from past mercenary spyware cases, which typically centered on a single malicious link sent to one target at a time.

How AdFection Works: Zero-Click Infection Through Ordinary Ads

The defining feature of AdFection is what it does not require. Citizen Lab’s analysis found that the attacks did not require a victim to click an advertisement, open a message, or click a link of any kind. An ad simply has to load on a page or inside an app the target happens to be using. From there, the ad platform’s targeting logic identifies the device, and a payload is delivered through the advertising pipeline itself rather than through a phishing message or malicious attachment.

Bill Marczak, a senior researcher at Citizen Lab, explained the mechanism in comments to Wired: “AdFection allows operators to automatically hack specific individuals with mercenary spyware as they use apps or browse the internet.” That framing matters because it shifts the threat model away from the user’s own judgment. Training people not to click suspicious links has been the baseline defense against spyware for a decade. AdFection sidesteps that defense entirely by not needing a click at all.

Marczak was equally direct about why the method is so hard to catch after the fact. “Someone who gets hacked like this would not see anything amiss,” he told Wired. “In fact, they might not even see any ad at all.” That absence of evidence, no suspicious message in an inbox, no unfamiliar app installed, no ad even visibly rendered, is what makes zero-click advertising attacks so difficult for both victims and forensic investigators to piece together after the fact.

Discover-Tech and the Controlium Platform

At the center of the report is Discover-Tech, an Israeli company whose advertising platform Citizen Lab says was used to target devices with spyware. The company’s product, called Controlium, functions on the surface like any other programmatic ad-serving system: it places ads across a network of partner sites and apps, profiles the devices that load them, and optimizes delivery. According to the reporting cited by CyberScoop, Controlium’s targeting layer went further than conventional ad-tech, enabling the spyware operators working with Discover-Tech to de-anonymize, fingerprint, and automatically infect devices the moment an advertisement rendered.

That description places Controlium in an unusual category. Most ad-tech fingerprinting exists to serve more relevant ads or detect fraud. Citizen Lab’s report describes a platform whose fingerprinting fed directly into a spyware delivery decision, raising the question of how many other ad platforms carry equivalent targeting precision without the same misuse. The report does not allege that mainstream ad exchanges knowingly participated in distributing the payloads, and Citizen Lab has not established that ad networks such as Google or Yandex were aware their infrastructure carried this traffic. What the lab does say is that those networks were identified as part of the distribution infrastructure that carried Discover-Tech’s ads to their final destinations.

Candiru and Bindecy: Two Spyware Makers, One Ad Pipeline

Citizen Lab linked the operation to spyware developed by Candiru, an Israeli vendor whose tools have surfaced in previous surveillance investigations. The lab also identified a second company, Bindecy, as another zero-click spyware maker that used Discover-Tech’s platform to reach targets. The presence of two separate spyware vendors on the same ad-delivery pipeline suggests Discover-Tech’s platform functioned less like a single vendor’s private tool and more like shared infrastructure that multiple surveillance firms could plug into.

That arrangement echoes a pattern already visible in other corners of the spyware industry, where a handful of specialized infrastructure providers quietly support several front-facing spyware brands. It also complicates any policy response. Sanctioning or blacklisting Candiru alone, as governments have done with mercenary spyware vendors before, would not necessarily disrupt Bindecy’s access to the same ad-based delivery channel, since the vulnerability sits in the advertising layer rather than in either company’s own code.

The Al Jazeera Infections: 45 Devices Confirmed

The most concrete casualty count in the report involves Al Jazeera. Citizen Lab says it linked Discover-Tech-delivered Candiru spyware to 45 Windows devices within the news organization during 2024 and 2025, a finding it reached with high confidence. The Citizen Lab report states plainly: “We call the technique AdFection, and show how it was used to infect devices at Al Jazeera with Candiru’s spyware.”

The Globe and Mail noted that the Al Jazeera case gives the investigation a concrete, named victim organization rather than an abstract technical finding, which is part of why the story moved so quickly through international outlets on the day of publication. Journalists have been a recurring target of mercenary spyware for years, from Pegasus infections documented in 2021 to more recent cases, and the Al Jazeera disclosure fits that pattern while introducing a delivery method that did not previously have public confirmation.

Citizen Lab was careful to note what it could not confirm. The number of people whose devices were actually infected beyond the 45 documented Al Jazeera cases remains unknown, and the lab said it was unclear which governments were behind the attacks. Those gaps are significant: a campaign that reached dozens of countries and hundreds of thousands of sites almost certainly extended well beyond one newsroom, but Citizen Lab’s public findings stop at what its forensic evidence can support.

Scale of the Campaign: Four Billion Records, 200,000 Sites

The numbers in the report are what separate AdFection from a typical targeted-hacking disclosure. Reviewing four billion advertising records required Citizen Lab to work at a scale closer to an ad-fraud investigation than a traditional malware teardown. Discover-Tech’s ads reportedly ran across close to 200,000 websites and mobile apps over a four-month window, appearing in dozens of countries rather than a single region.

That breadth does not mean every person who loaded one of those ads was infected. Citizen Lab has not established that every device reached by the advertisements was compromised, and the lab’s own findings describe targeting logic designed to select specific individuals rather than infect indiscriminately. The four billion records and 200,000 sites describe the delivery surface available to the platform, not a confirmed casualty count. Still, the sheer size of that surface is what alarmed researchers: a targeting system embedded that deeply in ordinary web and app advertising means the infrastructure for this kind of attack is already running through infrastructure millions of people pass through every day.

Why Governments Still Can’t Be Named

One of the report’s more frustrating findings for readers looking for accountability is that Citizen Lab said it was unclear which governments were behind the attacks it documented. Attribution in spyware cases is notoriously difficult. Forensic researchers can often trace a payload to a vendor, and sometimes to the client-management infrastructure a vendor uses to separate different government customers, but proving which specific agency pulled the trigger on a specific target usually requires either a vendor’s internal records or a victim’s own country-level investigation.

Politico reported that Candiru itself was rebranded as Integrity Labs in 2025, adding another layer of corporate reshuffling that can obscure which entity is accountable for spyware deployed under the Candiru name during the period Citizen Lab investigated. Corporate restructuring of this kind is a recurring feature of the mercenary spyware market, where rebrands and acquisitions have previously made it harder for journalists, researchers and regulators to keep a consistent paper trail on any single vendor.

Historical Context: From Pegasus Links to Invisible Ads

Citizen Lab has been the primary research body exposing mercenary spyware for more than a decade, with its Pegasus investigations putting NSO Group under international scrutiny starting in 2016 and intensifying through 2021. Those cases typically involved a malicious link sent over SMS or a messaging app, requiring at least some interaction, even if that interaction was as simple as the phone receiving a message that triggered an exploit automatically. The industry has steadily moved toward attacks that need zero user interaction at all, and AdFection represents the next step in that progression: instead of exploiting a messaging app’s rendering pipeline, it exploits the advertising pipeline that nearly every app and website already depends on for revenue.

That shift mirrors a broader pattern across the security field this year, where attackers increasingly target the infrastructure layers that regular users never see, rather than tricking users directly. The French supermarket refrigeration attacks disclosed earlier in 2026 followed a similar logic, compromising infrastructure that stores depended on rather than targeting employees’ inboxes. Supply-chain and infrastructure attacks have displaced phishing as the preferred entry point in several of this year’s largest incidents, including the hardware-level compromise Ledger confirmed in its hardware implant disclosure.

Market Impact: Ad Tech Faces a Trust Problem

The immediate market question is how much scrutiny mainstream advertising exchanges will now face. Programmatic advertising already runs on real-time bidding systems that pass device and location data between dozens of intermediaries in the milliseconds before an ad loads. Citizen Lab’s report demonstrates that at least one platform built on that same basic architecture was repurposed for spyware delivery, which puts pressure on every ad exchange to show it has controls that would catch similar abuse.

Publishers and app developers who unknowingly carried Discover-Tech’s ads across their inventory face a reputational exposure they did not create. Ad verification and brand-safety vendors, whose entire business is scanning ad creative and destinations for malware and fraud, will likely see renewed demand from publishers wanting assurance that their ad slots cannot be weaponized the same way. Expect ad-tech security audits to become a bigger line item for large publishers in the coming budget cycle, following a pattern similar to how the TFG cyberattack pushed retail brands to re-examine third-party vendor access across their networks.

Competitive Comparison: Spyware Delivery Methods

AdFection is best understood against the other delivery methods mercenary spyware vendors have relied on over the past several years. The table below compares the main approaches by interaction required, detection difficulty and typical target profile.

Delivery method	User interaction needed	Detection difficulty	Typical target
Phishing link (SMS or email)	Click required	Moderate: link or message can be forensically recovered	Individuals with known contact details
Zero-click messaging exploit (Pegasus-era)	None, message delivery triggers exploit	High: requires recovering the triggering message	Journalists, activists, officials with known numbers
Network injection (man-in-the-middle)	None, requires network position	High: needs network-level monitoring to catch	Targets on a known network or ISP
AdFection (ad-based zero-click)	None, loading any ad can trigger it	Very high: no message, file or visible ad may remain	Anyone the ad platform fingerprints as a match, no known contact needed

The key difference in the bottom row is targeting without a known contact point. Earlier zero-click methods still needed the operator to have a phone number or network position tied to the target. AdFection’s reliance on ad-platform fingerprinting means a target can be identified and hit purely through behavioral and device signals the advertising ecosystem already collects for commercial purposes, which is a lower bar for an operator to clear than acquiring a working phone number.

Report at a Glance: The Numbers Behind AdFection

Metric	Figure reported
Report length	72 pages
Publication date	October 15, 2026
Ad records analyzed	Four billion
Sites and apps carrying the ads	Nearly 200,000
Campaign window measured	Four months
Countries where ads appeared	Dozens
Confirmed infected devices (Al Jazeera)	45 Windows devices
Years covered by confirmed infections	2024 and 2025
Spyware vendors identified	Candiru and Bindecy
Ad platform identified	Controlium, operated by Discover-Tech

Expert Reactions to the Report

Citizen Lab’s own framing of the discovery leaves little room for ambiguity about how the lab views the significance of what it found. In the report itself, the lab states: “This is the first-ever public confirmation of zero-click mercenary spyware delivered via targeted ads.” That line, repeated across the lab’s research page, is the core claim the rest of the report’s technical evidence is built to support.

On the mechanics of Discover-Tech’s platform, the report adds detail that goes beyond simple ad targeting, describing Controlium as a system that enables its mercenary surveillance partners to de-anonymize, fingerprint, and automatically infect devices the moment an advertisement is rendered. That description, drawn from the report’s own technical findings, is what turns Controlium from a conventional ad-tech platform into what Citizen Lab characterizes as infrastructure built for surveillance clients.

Marczak’s comments to Wired round out the picture from the researcher who worked the technical side of the investigation, emphasizing both the automation of the targeting and the near-total absence of evidence left behind. Taken together, the lab’s public statements paint a campaign that was built to be scalable, automated and quiet, three qualities that made it difficult to catch even as it ran across hundreds of thousands of ad placements for months.

What Happens Next: Five Predictions

Regulators in the European Union and United States will likely open inquiries into Discover-Tech and the ad exchanges that carried its inventory, mirroring the scrutiny that followed earlier Citizen Lab Pegasus disclosures.
Browser makers and mobile operating system vendors will probably accelerate work on ad-rendering sandboxing, since AdFection’s exploit path runs through the same ad-rendering code every browser and app relies on daily.
Ad verification and brand-safety companies will see a near-term surge in publisher demand as site owners seek assurance their ad inventory cannot be used for similar targeting.
Additional victim organizations beyond Al Jazeera are likely to surface in the coming weeks as other newsrooms and civil society groups request their own forensic reviews from Citizen Lab or similar labs.
Expect continued uncertainty over government attribution, since Citizen Lab’s own report leaves that question open and spyware vendors rarely disclose their government clients voluntarily.

The Bigger Picture for Everyday Internet Users

Most people are not high-value targets for a mercenary spyware operation, and AdFection’s victim profile so far centers on journalists and the organizations that employ them. But the report matters beyond that narrow group because of what it confirms about the advertising ecosystem everyone uses. Ad-tech fingerprinting, the kind of device profiling that normally decides which sneaker ad or travel deal shows up in your feed, has now been shown in a documented case to double as targeting infrastructure for spyware.

That overlap raises uncomfortable questions for an industry that has spent years defending granular ad targeting as a harmless commercial practice. It also reinforces a trend security teams have been tracking all year: attackers increasingly go after the shared infrastructure layer rather than the individual, a dynamic also visible in the credential exposure behind the Denmark password breach and in the law enforcement case against the second ShinyHunters suspect, where stolen access credentials rather than individual phishing lures drove the scale of the damage.

For organizations handling sensitive data or working with at-risk sources, like newsrooms, the practical takeaway is that endpoint monitoring now needs to account for infection vectors that never touch a messaging app or email client at all. Security teams reviewing their own exposure might look at how they vet third-party code running in browsers and apps, an area that has already drawn regulatory attention in cases like the OpenAI bug case, where the handling of a security flaw became a legal matter rather than a quiet patch.

Where This Leaves Government AI and Surveillance Oversight

The AdFection disclosure also lands at a moment when governments are already under pressure to explain how they oversee advanced digital tools before those tools reach the public. The UK’s recent review flagging risks in Google’s Gemini 4 Argon model, detailed in the AI Security Institute’s assessment, shows regulators are already trying to get ahead of new technical capabilities before deployment. Mercenary spyware has historically moved faster than that kind of oversight, since vendors like Candiru and Bindecy operate as private contractors selling to government clients rather than as public companies subject to the same disclosure requirements as AI labs.

Citizen Lab’s report does not call for specific legislation, but the scale of the four billion records reviewed and the near-invisible nature of the attack method make a strong case for export-control style restrictions on ad-tech platforms that offer device fingerprinting precise enough to support this kind of targeting. Whether any single country moves first on that front, or whether it takes a coordinated response similar to the one that eventually targeted NSO Group, remains an open question the report itself does not answer.

Frequently Asked Questions

What is AdFection?
AdFection is the name Citizen Lab gave to a technique that delivers mercenary spyware through ordinary online advertisements, without requiring the target to click the ad or interact with it in any way.

Who published the AdFection report?
Citizen Lab, a research lab at the University of Toronto’s Munk School, published the 72-page report on October 15, 2026.

What is Discover-Tech?
Discover-Tech is an Israeli advertising company whose platform, Controlium, Citizen Lab says was used to identify and target devices for spyware infection through ad delivery.

Which spyware companies were named in the report?
Citizen Lab linked the campaign to spyware from Candiru, and separately identified Bindecy as another spyware maker that used Discover-Tech’s platform.

Were any organizations confirmed as victims?
Yes. Citizen Lab said it linked Discover-Tech-delivered Candiru spyware to 45 Windows devices within Al Jazeera during 2024 and 2025, with high confidence.

Do I need to click an ad to be infected?
No. Citizen Lab’s findings indicate the attacks did not require a victim to click an advertisement, open a message, or click a link, which is what makes the method difficult to detect.

Do we know which governments were behind the attacks?
No. Citizen Lab said it was unclear which governments were behind the attacks documented in the report, and that detail remains unconfirmed.

Were Google or other major ad networks involved knowingly?
Citizen Lab identified major ad networks as part of the distribution infrastructure that carried the ads, but the report does not establish that those networks knowingly participated in the spyware operation.

Citizen Lab Exposes AdFection: Spyware Hit 45 Devices [2026]