ChatGPT, Claude, and Grok all hit service trouble at the same time on September 3, an unusually wide failure for the AI industry’s flagship assistants. Downdetector logged complaints across all three services, and users also reported issues with Google’s Gemini, though Google did not issue a confirmed outage notice.
OpenAI opened an incident on its status page at about 2:56 PM listing degraded performance across 19 components spanning ChatGPT and Codex, from Conversations, Login, and Search to Voice mode, Deep Research, Agent, and the Codex command-line interface. Its first update said only that the issue was being investigated.
The disruption did not touch OpenAI’s API group, which covers Chat Completions, Responses, Fine-tuning, Embeddings, and Realtime, and the Ads Platform and FedRAMP groups were also unaffected.
Outages are common, but it is rare for three leading assistants to wobble simultaneously, and the episode underscored how much traffic now rides on a small set of model providers. Anthropic and xAI had not posted matching incident reports, and the simultaneous failures revived questions about concentration risk in AI infrastructure.
