Researchers Oleg Zaytsev and Ofek Ronen at browser security company Island disclosed the operation on October 6 after tracking a network of sites pitched to advertising professionals as tools for campaign optimisation, spending audits and business-account connections. They said hundreds of victim submissions reached the platform during their investigation and activity was continuing when their findings were prepared.
The campaign differs from simpler credential-harvesting pages by presenting users with a plausible business service before requesting access. Fake products carrying the branding of OpenAI’s ChatGPT, Google’s Gemini, Anthropic’s Claude, Perplexity, Manus and Meta’s Muse were observed, although there is no indication that the legitimate companies were involved in the operation.
Each lure ultimately directs a visitor to a Connect button. Clicking it produces a Browser-in-the-Browser, or BitB, interface: a counterfeit browser window rendered within the genuine browser tab. The simulated window can display a lock icon and an address bar showing a trusted destination such as accounts. google. com even though the outer browser remains on an attacker-controlled domain.
Island said the fake interface adapts to Windows, macOS, iOS and Android and can imitate details including Safari’s URL display, Chrome custom tabs and dark mode. The technique is designed to defeat a common safety check in which users inspect the apparent address bar before entering credentials.
Behind the imitation login window, the platform operates as an interactive system controlled by a human attacker. Researchers found that it fingerprints a visitor’s device and can retain several password attempts. An operator can reject an entered password, ask the victim to try again and preserve each submitted value while attempting to access the genuine account separately.
The system also allows operators to determine which authentication challenge appears next. Its observed workflows supported Google, Meta, TikTok and Okta, including requests for SMS codes, authenticator-app codes, Google approval prompts and Okta approvals. Commands were relayed while the victim remained engaged, enabling the attacker to adapt the phishing sequence to the authentication response encountered during a live account takeover attempt.
That human involvement is significant because conventional multifactor authentication based on reusable one-time codes can still be defeated when a victim supplies the code immediately to an attacker who is simultaneously signing into the real service. Phishing-resistant methods such as passkeys and hardware security keys bind authentication more closely to the legitimate service and are designed to reduce that exposure.
The lures specifically use language familiar to advertising agencies, media buyers and administrators. Island found a fake ChatGPT product promising Google Ads briefings, a Gemini-themed service offering manager-account and linked-client support, a Claude advertising portal and a Perplexity-branded service promoting campaign planning and spending audits.
Advertising manager accounts are attractive targets because one compromised identity may provide access to several client accounts, stored payment methods and approved advertising budgets. Attackers gaining administrative control can potentially change account permissions or use existing advertising capacity for unauthorised campaigns.
The infrastructure was reused beyond AI-themed advertising pages. Island linked the same underlying technology to refund, billing and recruitment lures, and said exposed source code in misconfigured public GitHub repositories helped reveal common routes, password-retry logic and a Telegram-linked control channel.
One backend identified by the researchers appeared in 73 archived scans covering 25 page domains between May 27 and June 20. Those pages included AI advertising lures as well as refund and recruitment sites, indicating that different front-end stories were connected to common infrastructure.
Follow Arabian Post
Select Arabian Post as your preferred source on Google and MSN News for trusted business news and Arab politics and updates.
