Overview
- Multiple leading chatbots — OpenAI’s ChatGPT, Anthropic’s Claude, and xAI’s Grok, with reports of Google’s Gemini also affected — showed simultaneous outages or elevated error rates on Sept. 3, 2026 that disrupted core features for many users.
- OpenAI reported elevated errors across roughly 15 ChatGPT and Codex components, with engineers detecting problems around 10:58 AM ET on Sept. 3, 2026 and the main degradation running until about 2:56 PM ET before systems were restored.
- Anthropic and xAI acknowledged partial outages for Claude and Grok respectively, said they applied fixes and returned services to normal within hours, and Anthropic reported it had identified a cause for its elevated errors.
- Outage trackers logged tens of thousands of user reports for ChatGPT and thousands for other platforms, and the incidents temporarily disabled features such as logins, Work Mode, file uploads, image generation, APIs and developer flows that many businesses rely on.
- No provider has confirmed a shared root cause; speculation tying the failures to OpenAI’s upcoming Astra model or to cloud vendors such as Cloudflare or Microsoft Azure remains unproven, highlighting real operational risk for organizations that depend on a single AI provider.