Scammers used phishing sites with fake AI products and ad accounts to steal advertisers’ login credentials
Advertisers were targeted by fake ChatGPT, Claude, Gemini, and Perplexity sites in a campaign designed to steal information.
The sites used false AI models to steal login credentials and multi-factor authentication (MFA) codes.
This is a phishing operation which used the launch of the Meta‘s Muse AI agent – a personal assistant for business and personal tasks – to trick audiences.
Malicious fake pages disguised as AI sites were used to lure media buyers, agency staff, and administrators to enter their details with the intention of compromising accounts and potentially gain access to downstream clients.
Agency staff and administrator accounts can act like a point of entry to a network of client environments, so one compromised login could give attackers access to any connected account.
A single breach could put an entire client portfolio at risk.
The false campaign
The phishing sites used fake AI products to entice decision-makers by promising to help find media buyers, obtain ad briefs, manage adspend as well as plan and audit campaigns.
Once attackers gained access, they could spend any outstanding balances on fraudulent campaigns, or sell the information on the dark-web.
Arturas Bubokas, cybersecurity expert at NordLayer, told Affiliate Leaders: “Attackers can spend available balances on fraudulent ad campaigns or resell the accounts to other cybercriminals for significant sums.
“The stolen credentials include multi-factor authentication (MFA) codes, meaning the attacker has full, authenticated access to the account.”
Users were prompted to connect their ad accounts to the fake AI product. But, the ‘connect’ button linked to an illegitimate Google window within the page, which even showed an address bar with the url ‘accounts.google.com.’
Protecting the business and clients
This technique, browser-in-the-browser, is not new. It is particularly effective because the fake window, which is styled as a login pop-up, is very realistic and in-line with the url the audience might expect.
Advertisers are most at risk as the infrastructure of the cybercriminal campaign is much wider than just a few websites.
Researchers Oleg Zaytsev and Ofek Ronen found that cybercriminals operated a multitude of sites including fake refund and recruitment pages.
Bubokas continued: “Beyond the account owner experiencing financial loss, a hijacked ad account can be used to push malicious or scam ads to real audiences under a trusted brand’s name, damaging the advertiser’s reputation and potentially triggering account bans from the platform.”
To best protect themselves from these attacks, advertisers should use browser-level threat blocking tools like an anti-malware extension. These will block phishing sites, and limit the attack surface.
Bubokas added that advertisers should also use a phishing-resistant MFA: “This fake AI platform phishing campaign works by capturing SMS and authenticator codes in real-time. Hardware keys and passkeys don’t use code at all. Instead, they verify that the login origin matches the service where the key was registered, and the browser refuses to release credentials to a phishing page, preventing authentication entirely. ”
Keeping good cyber hygiene is key, and exercising caution with links, pop ups, and websites that are unfamiliar will go a long way to protect you and your brand.
