Overlapping AI outages need not have one cause: OpenAI's 3 September 2026 routing error

Original: OpenAI’s incident write-up for 3 September 2026 (approximately 07:43–08:20 PDT). A routing configuration update was incorrectly interpreted and stopped requests reaching healthy services; rollback restored traffic, and returning load extended recovery. Affected surfaces named in its report: ChatGPT web/mobile/desktop and Codex CLI/desktop, plus some sign-ins, account creation and published sites. This report does not claim ChatGPT API inference was down; do not label an API outage from ChatGPT symptoms.

News and user reports called simultaneous Claude, ChatGPT and Grok failures a shared-cloud outage. A temporal cluster alone cannot show a shared supplier; OpenAI attributes its own episode to internal routing, not Azure, Cloudflare or Anthropic, and its postmortem makes no claim about the causes at those firms. This is a partial rebuttal to causal inference from timestamps, not proof the outages were wholly unrelated. Compare the documented June 2025 KV → Workers AI pathway and dependency test. Next discriminating test for 3 September: other providers’ original postmortems with dependency edges or incident IDs, not correlated Downdetector curves.