OpenAI permanently bans 2-year Plus subscriber over OAuth hijack, emails lost to spam

A long-time ChatGPT Plus subscriber—paying $20/month for over two years and using the service daily for development work—was permanently banned by OpenAI for 'Cyber Abuse' (phishing/malware under the TOS). The user insists no such activity originated from them and suspects their OAuth token (used with OpenClaw on a new VPS) was hijacked around May 7. Instead of detecting the breach and protecting the account, OpenAI sent four separate warning emails over six days. All four emails—three 'Usage Policy Violation' warnings and one 'Access Deactivated' notice—were automatically filtered into Outlook's spam folder. The user never saw them until being forcibly logged out on May 13.

The appeal process was equally troubling. An initial appeal via the help center was rejected within an hour, with OpenAI stating the decision was final and no further appeals would be considered. A follow-up human response offered no specifics—no example prompts, no IP addresses, no geolocation data—citing 'security and privacy reasons.' The user, a tech professional, argues that in-app notifications, pop-ups, or even a single blocked prompt would have alerted them. They also note that OpenAI's own documentation admits emails often hit spam filters. The incident raises serious questions about how the company handles account integrity, breach detection, and customer support for long-term paying users.