Por Renan Filho · 2026-10-17**Lead**
A growing wave of cyberattacks is leveraging Google's advertising network to trick users into downloading fake versions of popular artificial intelligence applications, security researchers have warned. The malicious ads appear at the top of search results for widely used AI tools such as ChatGPT, Midjourney, and Sora, luring users who are eager to access or download these platforms. Once clicked, the ads redirect victims to fraudulent websites that either infect devices with information-stealing malware or prompt the installation of seemingly legitimate but compromised software. The campaign has been observed in the United States and is expected to expand as cybercriminals refine their techniques.
**Context**
The scheme works by taking advantage of Google's ad auction system, where threat actors create ad copy that closely mimics official download pages or product pages for AI services. Because Google's automated review processes cannot always catch subtle but dangerous differences, these malicious ads are often served to users searching for keywords like "ChatGPT download," "Midjourney free," or "AI image generator." The landing pages frequently use official logos, familiar color schemes, and near-identical domain names — for instance, replacing a letter or using a common misspelling — to reduce suspicion.
Once a user downloads what they believe is a legitimate installer, the payload executes a multi-stage infection. Common malware families observed include infostealers such as RedLine, Vidar, and Lumma, which are designed to harvest browser passwords, cryptocurrency wallets, credit card data, and other sensitive information. In some cases, the malware also deploys remote access trojans (RATs) that allow attackers to take control of the victim's machine. The attacks have been reported by multiple cybersecurity firms, who note that the campaigns are particularly effective because they bypass traditional spam filters and rely on users' trust in Google's ad system.
**Analysis**
The rapid rise of generative AI tools has created a fertile environment for social engineering. As millions of Americans rush to experiment with or adopt AI for work and personal use, the gap between genuine demand and supply has been exploited by criminals. Users who are unfamiliar with official distribution channels — such as directly visiting the OpenAI website for ChatGPT or using the official app stores — are especially vulnerable. The use of Google Ads gives the attacks a veneer of legitimacy, making it harder for even cautious users to distinguish between a real advertisement and a malicious one.
This attack vector also highlights a broader security challenge: the tension between advertising revenue and user safety. Google has faced repeated criticism for allowing malicious ads to slip through its verification process, though the company has stated it removes violating ads and bans bad actors when identified. However, the speed at which new campaigns are launched often outstrips takedown efforts. For U.S. businesses and individuals, the consequences can be severe — beyond immediate data theft, credentials harvested in these attacks can be used for follow-on intrusions, corporate espionage, or ransomware deployment. The trend is not limited to AI tools; similar campaigns have targeted software like Zoom, Slack, and Notion. Yet the AI boom provides a unique opportunity for attackers because the technology is both novel and in high demand.
**Fontes**
This article is based on reporting from cybersecurity researchers at Malwarebytes, SentinelOne, and BleepingComputer, who have documented multiple waves of malicious Google Ads targeting AI users in 2025 and 2026. Additional context drawn from public advisories by the Cybersecurity and Infrastructure Security Agency (CISA) regarding malvertising trends affecting U.S. internet users.