ChatGPT, Claude and Grok Fell Together, and Nobody Knows Why

2 min read
All blog articles
ChatGPT, Claude and Grok Fell Together, and Nobody Knows Why
37 000+ Downdetector reports against OpenAI
3 major AI assistants hit
~2 h approximate outage duration

At 9 a.m. Eastern this Thursday, three fierce rivals finally agreed on something: none of them would answer you. ChatGPT, Claude and Grok all went dark within minutes of each other.

ChatGPT, Claude and Grok Down at the Same Time

Downdetector lit up just after 9 a.m. ET. By 11 a.m., users had filed more than 37,000 reports against OpenAI alone.

On ChatGPT, the damage covered conversations, logins, file uploads, Voice, Search and Deep Research, plus Codex on the developer side. Anthropic flagged elevated errors on Claude, while xAI confirmed a Grok incident.

Cursor and Gemini Got Dragged In Too

Cursor went down as collateral damage, since the code editor leans on Claude and Grok under the hood. One upstream failure, one silent editor.

Gemini, now close to a billion users, saw its own spike in reports. Google never declared an official outage, though.

No Confirmed Cause for the AI Outage

None of the three companies has explained what happened. Observers noticed simultaneous hiccups at Cloudflare and AWS, but nothing ties them to the incident so far.

The scene echoes June 2024, when ChatGPT, Claude and Perplexity failed on the same day. The timing adds a twist: OpenAI teased news around Astra, its next big model, that very morning.

The Outage Ended, the Lesson Should Not

Services started to recover before 9 a.m. Pacific after OpenAI applied a fix. At Anthropic, the Opus models took longer than the rest to come back.

For businesses, the takeaway is blunt: one AI provider equals one single point of failure. Multi-model redundancy just earned its best sales pitch of the year.

Why did ChatGPT, Claude and Grok go down at the same time?

No shared cause has been confirmed. Cloudflare and AWS showed issues in the same window, but OpenAI, Anthropic and xAI each run their own separate investigation.

How long did the September 3 AI outage last?

Roughly two hours for most users. Reports surged just after 9 a.m. ET, and the major services answered normally again by early afternoon.

Who Went Down on September 3, 2026

ServiceStatus during the incident
ChatGPTConfirmed outage: chat, Voice, files, Codex
ClaudeElevated errors, Opus models slower to recover
GrokIncident confirmed by xAI
GeminiReport spike, no official outage
CursorUnavailable, relies on Claude and Grok
EL
Emma Lawson Emma Lawson covers AI regulation, policy shifts, and their impact on the tech industry for AIxploria.