Overview
- Multiple leading chatbots including OpenAI’s ChatGPT and Codex, Anthropic’s Claude, and xAI’s Grok reported elevated errors or downtime on Thursday, September 3, 2026, with user reports spiking on outage trackers.
- The outages disrupted core features such as conversations, login, file uploads, image generation, voice mode, developer APIs, and ChatGPT Work, affecting both casual users and paying enterprise customers.
- Companies posted status updates, applied mitigations or fixes during the day, and restored service within hours with OpenAI marking its incident resolved after several hours and Anthropic and xAI reporting fixes for Claude and Grok respectively.
- No single, publicly confirmed common root cause emerged from provider updates or reporting, though xAI/SpaceX attributed Grok’s disruption to an outage at its Memphis compute center and speculation about third‑party/cloud providers circulated.
- The episode underlines a pattern of brief but visible AI outages in 2026 and is prompting businesses to reassess redundancy plans, SLAs, and the technical difficulty of building true multi‑provider failover for model‑dependent workflows.