Malware blitz: ads become industrial-scale hack delivery

The early October 2026 wave did not appear out of nowhere; it followed an already proven playbook in which ad-borne malware campaigns combined redirects, impersonation, and selective delivery to move different payloads through mainstream web traffic. Before August 2025, attackers used malvertising involving redirects, brand impersonation, and cloaking to deliver malware by showing benign content to reviewers but malicious content to users, a precursor that helps explain how coordinated operations could simultaneously exploit Google Ads, Bing redirects, and Meta ads while swapping in newer malware families and infection chains.

What made those October 2026 campaigns especially adaptable was the maturation of ClickFix as a delivery method that could sit inside the same ad-driven infrastructure while pushing victims toward hands-on compromise. Microsoft security researchers described ClickFix on August 21, 2025, as “a social-engineering method persuading users to execute malicious commands,” noting it “replac[ed] earlier automatic downloads,” and the technique had already been tied to real malware because “a campaign using ClickFix delivering Lumma Stealer was observed in April 2025,” showing that diverse payloads were already being operationalized before StreamRat-era activity emerged.

Ad platforms work because they lend attackers borrowed credibility at the exact moment users are looking for a trusted product. As IBM Technology put it, “That ChatGPT ad might not be from OpenAI,” describing how attackers bought sponsored search ads that led to a fake “Plus 5.6” Custom GPT, then funneled users onward; Bleeping Computer similarly tied Google Ads and Bing redirects to Claude-themed ClickFix delivery, showing that the ad itself is not just bait but the first operational layer in the malware chain, one that exploits the trust and reach of major ad platforms before victims ever see the payload.

What makes that delivery layer effective is the combination of redirects, impersonation, and delayed malicious action that helps campaigns slip past both users and defenses. In the ChatGPT case, the “sponsored search ads that led to a fake ‘Plus 5.6’ Custom GPT” then “steered at least 40 victims toward a ClickFix page and a remote access trojan,” while the broader tactic is captured by IBM Technology’s question, “How do you catch an imposter that looks exactly like your security tools?”—posed in discussing how “Rapid7 found Linux backdoors… hiding on the appliances that screen companies’ email”—a reminder that cloaking works by making the dangerous step appear routine until the victim triggers it.

The evidence now points to industrialized abuse, not scattered experiments. Ars Technica reported the headline “Google Ads Scam Locks Browsers; Netskope Flags 250 Campaigns, Calls for Immediate Action,” and that Netskope “Flags 250 Campaigns” tied to the operation, while Citizen Lab separately “sifted through nearly four billion ad-bidding records” and found Discover-Tech ads on “more than 70,000 mobile apps and 120,000 websites between March and June 2026,” showing a footprint measured in hundreds of campaigns and exposure across a vast slice of the ad-supported internet.

That scale is also visible in who was reached and on what devices. Bleeping Computer described Group-IB uncovering the RemControl Android banking trojan hijacking TVTap “Across Europe and Canada,” while The Register reported Meta ads steering Polish Android users into a premium-rate billing trap; in the higher-end espionage tier, Citizen Lab said it “traced attacks against at least 45 devices at Al Jazeera in 2024 and 2025 to Candiru spyware delivered through ads from Discover-Tech,” with the same ad presence appearing on networks “belonging to government agencies, defense organizations, and NATO.”