SME Cybersecurity: ChatGPT Phishing Targets Logins and Payment Detail
Cofense has identified phishing that impersonates ChatGPT payment notifications to steal login credentials and payment information. For UK SMEs, the risk sits where AI adoption meets routine subscription management. Verify billing through the official service, protect accounts with multi-factor authentication, and give staff a clear reporting route. If information has been disclosed, act immediately.
A request to update a ChatGPT payment method can look like routine administration, yet hand an attacker the information needed to target an employee’s account and business card.
Cofense’s analysis, published on 17 September 2026, describes attackers impersonating ChatGPT billing notifications to obtain credentials and payment information. The underlying trick is familiar: make an unexpected message feel like an ordinary subscription task.
For UK SMEs adopting AI tools, the lesson is to bring account ownership and payment verification into everyday security procedures.
SME Cybersecurity: how does ChatGPT billing phishing work?
The attack exploits trust in a familiar service rather than requiring a weakness in ChatGPT itself.
Why would a payment notification persuade an employee?
Phishing is a deceptive message designed to make someone disclose information or perform an unsafe action. Credentials are the information used to access an account, typically an email address and password.
Cofense describes fake notifications claiming that a payment method needs updating. Subscription billing gives attackers a credible business context: recipients may worry about losing access to a tool they use daily.
An owner-manager preparing proposals could act quickly to preserve access. An administrator managing several software subscriptions might treat the request as another payment task. The mitigation is the same: check the account through the official service, independently of the email.
Does this mean ChatGPT has been breached?
No. Impersonating a service is not evidence that its systems have been compromised. Cofense’s report concerns phishing targeting users, not proof of a breach of OpenAI.
The distinction matters when communicating with staff. Describe the risk as a fraudulent billing message, not an announcement that ChatGPT is unsafe.
The wider UK context supports taking this seriously. The government’s Cyber Security Breaches Survey 2025 found that 85% of businesses identifying a breach or attack reported phishing. That statistic covers phishing generally, not this campaign or all UK businesses.
Why are AI subscriptions a business risk?
An AI account can hold business information as well as provide access to a paid service.
What could an attacker obtain?
A compromised account may expose conversations, uploaded files or other information accessible within that account. The impact depends on its configuration and what employees have stored there. Cofense’s findings should not be interpreted as proof that these consequences occurred in every case.
For a consultancy, prompts might contain draft proposals or client context. Limit unnecessary sensitive uploads and use approved accounts with clear ownership.
Payment information creates a separate fraud risk. Keep subscription billing with a named owner and require independent verification before changing card details.
How does unmanaged AI use increase exposure?
Shadow AI means employees using AI services without business approval or oversight. It leaves directors unsure which accounts exist, who pays for them and what information they contain.
Create a simple register of approved AI services, account owners and billing contacts. Avoid shared passwords, which make access control and incident investigation harder.
.jpg)
