Hackers Exploit ChatGPT’s Popularity With Fake Billing Notices That Target Credit Cards and OTPs

Hackers are impersonating OpenAI in fake ChatGPT Plus billing emails that lead users to payment pages designed to steal credit card details and one-time passwords.

The campaign combines convincing subscription warnings with a live operator who can decide when to request a verification code, potentially helping approve an unauthorized payment.

The emails claim a subscription payment failed and urge recipients to update their billing information. This pressure mirrors other AI brand impersonation campaigns that exploit trust in popular services to collect financial information.

In this operation, the checkout page keeps victims waiting while an attacker controls what appears next. Researchers from Fortra identified the phishing operation and documented its infrastructure in research published on October 26, 2026.

Importantly, they observed phishing, not malware deployment: no executable download or compromise of victims’ computers was seen during the investigation.

Hackers Exploit ChatGPT’s Popularity With Fake Billing Notices

The attack begins with an email posing as an OpenAI payment notice. Links pass through tracking services or redirect wrappers before reaching a fake checkout hosted on short-lived WordPress instances.

Researchers found the same landing page reused across multiple recipients within a sending wave, suggesting shared infrastructure behind separately tracked messages.

The checkout imitates a payment form, recognizing card brands and formatting card numbers, expiration dates, and security codes as users type.

Similar WordPress payment phishing schemes have used convincing checkout interfaces to steal card data and verification codes, although Fortra did not establish a connection to those operations.

After submission, the page sends the card number, expiration date, security code, and cardholder name to the server. It then displays a loading animation indefinitely.

The captured code explicitly instructs the page to keep the spinner running rather than process a response or advance automatically.

Meanwhile, the browser repeatedly contacts an operator dashboard, maintaining the victim’s session and checking for instructions.

A human operator can redirect the victim to a one-time-password capture page, display an error, or end the session. One test lasting four minutes and three seconds generated 969 requests, mostly dashboard polling traffic.

This creates an opportunity to capture a bank verification code while the victim believes a subscription update is underway.

Comparable real time OTP interception appears in other payment-fraud kits, but the observed ChatGPT operation uses repeated polling to receive instructions. Capturing a code could help authorize fraud; it does not mean every submission produces a successful transaction.

Fortra found identical client-side code and a shared application identifier across independently deployed sites examined roughly a week apart.

Sites disappeared and were replaced during the investigation. Whether attackers purchased hosting accounts themselves or compromised existing customer sites remains unresolved.

The infrastructure also complicates investigation. Unexpected requests received generic errors, while a token-generating path appeared to influence access.

Researchers reported that interception-based analysis disrupted the session flow, whereas passive browser inspection captured the payment submission and polling behavior without breaking it.

For defenders, recurring page structures and dashboard traffic offer stronger detection opportunities than isolated addresses. Broader live financial phishing operations demonstrate why stolen verification codes need urgent attention.

Users should verify billing through their account rather than an email link, while organizations should investigate matching messages and report fraudulent hosting. Broad infrastructure blocks require care because legitimate customers share the platform.

Indicators of compromise (IoCs):-

Note: IP addresses and domains are intentionally Qdefanged (e.g., [.]) to prevent accidental resolution or hyperlinking. Re-fang only within controlled threat intelligence platforms such as MISP, VirusTotal, or your SIEM.

Give your SOC the alert context to stops incidents quickly. Integrate Threat Intel from 16K SOCs.

The post Hackers Exploit ChatGPT’s Popularity With Fake Billing Notices That Target Credit Cards and OTPs appeared first on Cyber Security News.