Coordinated Outages Across Three Major Providers

On September 3, 2026, OpenAI, Anthropic, and Google all reported service disruptions affecting their flagship AI products in overlapping time windows. The incidents disrupted consumer interfaces, developer APIs, image generation tools, and account creation flows across the three providers.

OpenAI’s ChatGPT and Codex Outage

OpenAI detected elevated error rates across ChatGPT and Codex at approximately 10:58 AM UTC on September 3, 2026. The outage affected 15 components on ChatGPT and 4 on Codex. OpenAI engineers applied mitigations within approximately 24 minutes of the incident, resolving it by 11:22 AM UTC.

In a separate incident the same morning, ChatGPT Work Mode experienced a disruption that lasted approximately 5 minutes before resolving on its own.

Anthropic’s Claude Outage

Anthropic’s Claude Sonnet 5 experienced elevated errors on September 2, 2026. The incident lasted approximately 25 minutes before engineers resolved it.

Google’s Gemini Disruption

Google’s Gemini experienced user-reported disruptions on September 3, 2026, with at least one detected issue taking roughly 2 hours to fully resolve.

Uptime Context and Multi-Provider Redundancy Challenges

Independent uptime monitoring services recorded that major AI providers maintained availability at or above 99.9% across 2026. However, at 99.9% uptime, a service can still be down for nearly 9 hours over the course of a year.

For organisations relying on multiple AI providers for redundancy, challenges emerge in implementation. A prompt tuned for GPT-4o does not always produce equivalent output from Claude Sonnet 5 or Gemini, meaning genuine multi-provider redundancy requires engineering investment beyond a second API key.

Source: Crypto Briefing