CYBERRAKSHAK LABS Β· RESEARCH #064

🚨 AdFection: Zero-Click Spyware Delivered Through Targeted Online Ads

Citizen Lab and Confiant describe AdFection, a targeted-advertising chain linked to Discover-Tech and Candiru that could deliver mercenary spyware without a click. This investigation separates confirmed reporting from unknown infection scope and sets out practical detection and response priorities.

By Vivek Kumar Β· Published 17 October 2026
RESEARCH#064
CATEGORYThreat Intelligence / Spyware / Ad-Tech Supply Chain
CRL ASSESSMENTHIGH
RESEARCH LEVELDeep Research
PUBLISHED2026-10-17
Source & social links:
LinkedIn Post β†— WhatsApp β†— YouTube β†—
How CyberRakshakLabs researches threats β†’
A device may enter a spyware delivery chain simply by rendering a targeted advertisement. The unusual part is not the familiar advert itself β€” it is the ad-tech infrastructure used to profile a device, decide whether it is a target, and redirect selected devices toward an exploit chain.
45+Al Jazeera devices reportedly targeted β€” not a claim that every device was infected
2024–25Period in which the targeting activity was observed
0 clicksNo ad click required for the reported targeted delivery path
Evidence boundary: Citizen Lab's public investigation describes a targeted campaign and links the ad-delivery role to Discover-Tech and the spyware chain to Candiru with high confidence. The target count is not an infection count. The government customer was not publicly identified in the sources reviewed, and the complete exploit chain is not published.
1. Executive Summary β€” What Is AdFection?

A Citizen Lab investigation published on 15 October 2026 describes a technique researchers call AdFection: targeted advertisements were used as a delivery path for Candiru mercenary spyware. Devices used at Al Jazeera Media Network were targeted during 2024–2025. The investigation links Discover-Tech's advertising infrastructure to targeting and the hand-off to spyware infrastructure.

The key distinction is between an ad impression, a targeting decision, and a successful infection. A broad advertising impression did not automatically mean a device was infected. The chain selected devices that matched targeting criteria; those selected devices could then be redirected through validation and exploit infrastructure.

Core assessment: online advertising can act as an initial delivery surface for highly targeted spyware. Traditional advice such as β€œdo not click suspicious links” is not enough to describe this threat, because a click was not required in the reported path.
2. How the Zero-Click Delivery Chain Works

The source investigation describes a multi-stage sequence. The exact exploit implementation is not fully public, so the stages below summarize the reported chain rather than reconstructing exploit code.

Ad renders→Device profiling→Target decision→Candiru validation→Exploit delivery→Spyware C2
StageReported actionDefensive relevance
1. Ad impressionAn advertisement renders in a website or mobile app and executes its associated script.Exposure can occur through normal browsing or app use; the user may not click.
2. Profiling and telemetryAdvertising infrastructure receives targeting and impression-related information, potentially including IP and advertising/browser identifiers.Ad-tech endpoints and identifier leakage become relevant telemetry sources.
3. SelectionA server-side decision determines whether the device matches a target profile; non-targets may receive no exploit stage.Most impressions must not be equated with successful exploitation.
4. Validation and redirectSelected devices are passed to validation and then exploit infrastructure attributed to Candiru.Correlate DNS, proxy, browser/app and endpoint events in sequence.
5. Implant and command-and-controlFor successfully compromised devices, spyware may establish control channels and collect sensitive device information.Preserve network and device artifacts for specialist forensic review.
Source-attributed AdFection sequence showing ad rendering, target selection, Candiru validation, exploit delivery and command-and-control stages
AdFection sequence reproduced from the figure supplied with the research material. Source attribution: Citizen Lab. Treat the depicted domains as source-reported indicators and validate them against the primary report before operational use.
3. Why the Advertising Supply Chain Matters

Programmatic advertising connects advertisers, demand-side platforms (DSPs), ad exchanges, supply-side platforms (SSPs), publishers and the browser or app showing the ad. The ecosystem relies on rapid decisions and data exchange. The investigation describes how this legitimate infrastructure was allegedly repurposed to profile and select devices for surveillance.

Advertising ecosystem diagram showing advertiser, demand-side platform, ad exchange, supply-side platform, publisher and user, with data and third-party tracking paths
Advertising supply-chain illustration from the supplied research document. It explains the surrounding ecosystem; it is not a forensic map of every organisation or transaction in this specific incident.

Scale does not equal infection

Advertising reach and profiling telemetry can be broad while an exploit hand-off remains highly selective.

Trust can be inherited

Content arriving through ordinary publisher/ad channels may bypass the user's suspicion of a direct malicious link.

For defenders, this expands the third-party risk model: ad networks, embedded scripts, real-time bidding flows, analytics endpoints and content delivery paths can be relevant to an incident even where the user never visited a suspicious URL intentionally.

4. What the Evidence Supports β€” and What It Does Not
FindingAssessmentBoundary
Citizen Lab published an AdFection investigationPublicly documentedPrimary report dated 15 October 2026.
Devices used at Al Jazeera were targeted during 2024–2025Reported by investigatorsPublic reporting describes at least 45 devices targeted; do not interpret this as 45 universally confirmed infections.
Discover-Tech provided the ad-delivery bridgeAttributed in the investigationVendor/infrastructure attribution from Citizen Lab and Confiant.
Candiru spyware was linked to the infection chainHigh-confidence attribution in the reportThis does not reveal the identity of the ultimate customer.
Every device that displayed an ad was infectedNot supportedThe chain included target selection and subsequent exploit stages.
Exact exploit code, full infection count and government customerNot established in sources reviewedDo not fill these gaps with speculation.

The distinction between targeted devices, devices that passed validation, and devices for which investigators confirmed compromise matters for accurate reporting and response prioritisation.

5. Timeline and Campaign Context
  • 2021: The United States added Candiru to its Entity List, citing malicious cyber activities.
  • 2024–2025: Citizen Lab reports targeting of devices used at Al Jazeera; the organisation's technology team observed unusual patterns across this period.
  • 15 October 2026: Citizen Lab published the AdFection investigation describing the Discover-Tech/Candiru connection.
  • 16 October 2026: BleepingComputer published a technical summary of the findings.

The campaign is notable because it puts ad-tech targeting and spyware delivery in one operational chain. It should not be treated as proof that every advertising platform is involved or that all users face the same level of targeting.

6. Threat Actor and Attribution Assessment

Discover-Tech

Described by investigators as the advertising technology intermediary used to profile/target devices and bridge ad delivery with surveillance infrastructure.

Candiru

A commercial mercenary-spyware vendor whose spyware is linked to the reported infection chain with high confidence in the public investigation.

The specific government customer or ultimate operator was not publicly identified in the sources reviewed. Attribution should stay with the vendor and infrastructure relationships actually documented; this article does not assign a country, agency or named customer.

7. Potential Data Exposure β€” Capabilities vs. Proven Collection

Public reporting describes spyware capabilities that may include access to files and documents, browser data such as cookies/bookmarks/saved credentials, communications, keystrokes, screenshots, and microphone or webcam input.

Do not overstate the evidence: reported capabilities are not proof that every data category was collected from every targeted device. Device-specific collection should be established from forensic findings and the primary investigation.

If credential or session theft is suspected, rotate credentials and revoke active sessions from a known-clean device after coordinating containment and evidence preservation with incident responders.

8. SOC Hunting β€” What to Look For

Hunting should focus on the ordered event pattern rather than a single generic indicator. Use current, validated indicators from Citizen Lab and Confiant's technical appendices; infrastructure may change, and an individual domain match requires context.

βœ“ DNS and proxy connections to ad-tech endpoints
βœ“ Unusual iframe or redirect activity during an ad impression
βœ“ Short time gaps between profiling, validation and exploit infrastructure
βœ“ Browser/app crashes or unexpected process activity around suspicious network events
βœ“ Mobile threat-detection and EDR alerts on high-risk devices
βœ“ Outbound connections consistent with the source-reported command-and-control sequence

Prioritise correlation across DNS, proxy, firewall, mobile telemetry, EDR, browser/app records and identity activity. A single domain, crash or battery symptom alone is not proof of infection.

9. Incident Response and DFIR Priorities
  1. Preserve browser/app telemetry, device logs, DNS records, network flows, endpoint artifacts and relevant identity logs before rebuilding or resetting a suspected device.
  2. Escalate high-risk suspected compromise to a specialist mobile/endpoint forensics team; targeted mercenary spyware may evade ordinary antivirus checks.
  3. Correlate the exposure window across related staff devices, shared accounts, identity sessions and sensitive data access.
  4. Validate network indicators against the primary report and your own context before deploying blocks, avoiding broad blocks based only on unverified lists.
  5. If credential theft is suspected, rotate credentials and revoke sessions from a known-clean device; preserve the evidence needed to confirm scope.
  6. Coordinate legal, privacy, executive and notification decisions against verified findings and applicable obligations.
10. Practical Protection for Organisations and High-Risk Users
βœ“ Keep the operating system, browser and applications fully patched.
βœ“ Use managed devices, least privilege and centralised security telemetry.
βœ“ Consider reputable content blockers with IT approval for high-risk roles.
βœ“ Reduce ad tracking and disable mobile advertising identifiers where supported and appropriate.
βœ“ Assess ad-tech and third-party script exposure for sensitive environments.
βœ“ Restrict app installation and review mobile security configuration.
βœ“ Establish specialist escalation paths for suspected spyware.
βœ“ Treat crashes or battery drain as leads to investigate, not proof of compromise.

These measures reduce exposure or improve detection, but none alone guarantees protection from a targeted zero-click spyware operation.

11. Indicators and Verification Notes

The figures supplied with this research include example infrastructure labels and a staged network sequence. This article does not recreate a complete IOC appendix. Before creating block rules or detections, retrieve the current indicator set directly from the primary Citizen Lab and Confiant publications, confirm defanging and context, and evaluate false-positive risk.

For published intelligence, separate each item into source-reported indicator, observed local event and analyst-assessed linkage. Do not treat an infrastructure match as standalone proof that a device is compromised.

12. ATT&CK Mapping β€” Why Attribution Should Wait for Evidence

The public summary alone is insufficient to assign a complete, victim-specific ATT&CK technique chain with confidence. A defender may investigate drive-by compromise, exploit delivery, collection and command-and-control hypotheses, but final mapping should be based on the primary technical report and local forensic evidence.

Document what was directly observed separately from what is inferred from the reported sequence. Avoid representing every possible capability as an observed technique in this incident.

13. Sources, Verification and Research Limitations

Scope note: this article summarises public reporting available as of 17 October 2026. It does not independently examine affected devices, assert a complete infection count, identify the ultimate customer, or prove that every targeted device was infected.

14. CyberRakshakLabs Assessment β€” Advertising Is an Attack Surface

CRL Assessment: HIGH. The reported chain challenges the assumption that a user must open a suspicious attachment or click a malicious link before a device can be exposed. For high-risk individuals and organisations, ad-tech infrastructure should be considered within third-party risk, endpoint visibility and incident-response planning.

The best operational response is measured: update devices, reduce unnecessary ad tracking where feasible, centralise telemetry, validate source-reported indicators, and route suspected cases to specialists. Do not turn the campaign into a claim that every ad is malicious or every targeted device was infected.

Key takeaway: AdFection shows how a normal-looking ad impression can become the first step in a targeted spyware chain. No click may be required β€” but a target decision and further stages still separate broad ad exposure from a successful infection. Validate the evidence, correlate the chain, and investigate suspected compromise with specialist forensics.
Research integrity note: Campaign details and infrastructure attribution are presented as reported by Citizen Lab and Confiant. The source-reported β€œ45+” figure describes devices targeted; it should not be interpreted as the number of universally confirmed infections. Capabilities, indicators and attribution must be checked against the linked primary reports.

Think Before You Click. Stay Aware. Stay Secure.

CyberRakshakLabs β€” Threat Intelligence for a Safer Tomorrow.