OpenAI Clarifies Codex Limits: Subscription-to-API Sharing Triggers Anti-Fraud Systems
OpenAI Codex and ChatGPT lead Thibault Sottiaux clarified that recent usage restrictions on Codex are targeting users who convert subscription services into API traffic using tools like "sub2api." These unauthorized sharing and reselling behaviors trigger OpenAI's anti-fraud risk control systems. This enforcement highlights OpenAI's strict stance against bypassing official API pricing models through subscription-sharing proxies, affecting developers who rely on unofficial API wrappers. It reinforces the boundaries between consumer subscriptions and commercial API usage. Users logging in via "Sign in With ChatGPT" on official clients or supported third-party open-source clients like Pi and OpenCode remain unaffected. However, using platforms like Sub2API to distribute subscription quotas via custom API keys will be flagged.
## BACKGROUND
OpenAI Codex is an AI-driven coding agent suite designed to automate software engineering tasks. Sub2API is an open-source AI API gateway that allows users to pool AI subscriptions (such as ChatGPT or Claude) and redistribute them as API traffic, which violates OpenAI's terms of service.