Yesterday morning, Thursday 3rd September, ChatGPT, Claude and Grok (maybe Gemini as well) all saw overlapping outages. OpenAI cited a routing error. Anthropic reported an infrastructure issue on several Claude models. Grok’s came from an outage at the Memphis compute center, which also hit some compute partners. All services were restored within a few hours. No shared external root cause has been confirmed.

Why did all the major AI systems go down at the same time? What are they not telling us?

First dropped: September 4, 2026 | Last modified: September 04, 2026