The September 3 ChatGPT, Grok and Claude outage affected three leading AI services during overlapping windows, disrupting chatbot access, developer tools and APIs for some users. All three services later recovered, but the available reporting does not establish one shared failure behind the incidents.
OpenAI told WIRED and The Register that a routing error beginning at 7:43 a.m. PT made ChatGPT and its Codex coding service unavailable for some users across platforms. The company said it implemented a solution around 8:17 a.m. PT and continued monitoring the recovery.
The Verge reported that ChatGPT users encountered errors with conversations, sign-ins, file uploads, voice mode, search, deep research and image generation. OpenAI's status page listed elevated errors affecting ChatGPT and Codex during the incident.
Anthropic reported a partial outage beginning at 6:23 a.m. PT, according to WIRED. The company marked it resolved at 9:16 a.m. PT after saying it had identified the cause and deployed a fix. An Anthropic spokesperson told The Register that an infrastructure issue affected Claude.ai, Claude Code, Claude Cowork and the Claude API, and said service was restored at 16:16 UTC.
xAI began investigating Grok problems at 6:30 a.m. PT, WIRED reported. The incident affected Grok on web, Android and iOS, and xAI marked it complete at 10:05 a.m. PT, saying traffic was healthy again. SpaceX, xAI's parent company, attributed Grok's disruption to an outage at its Memphis compute center.
Were the ChatGPT, Grok and Claude outages connected?
There is no confirmation that they were. The failures overlapped in time, which can suggest a common dependency such as a cloud provider or content-delivery service, but timing alone is not evidence of a common root cause.
OpenAI named a routing error. Anthropic described an infrastructure issue without publicly identifying a shared external provider in the reporting reviewed by WIRED and The Register. SpaceX tied Grok's problem to its own Memphis compute center. Those statements explain parts of each incident, not the relationship between them.
Cloudflare told The Register that it had no significant service disruption. The status pages for Amazon Web Services, Google Cloud and Microsoft Azure also showed no relevant technical difficulties at the time, according to The Register. WIRED likewise reported that Cloudflare, AWS and Azure had not reported outages. Clean public dashboards are not a complete map of every dependency, but they do not supply evidence of a broad provider outage here.
What the outage affected
- ChatGPT and Codex: Some users could not access the services while OpenAI addressed a routing error.
- Claude: Anthropic's partial outage reached its consumer chatbot, coding tools, Cowork service and API.
- Grok: xAI reported cross-platform problems and later linked them to its Memphis compute center.
The incident was an overlap, not a demonstrated coordinated event, cyberattack or internet-wide outage. The available reporting does not tie the failures together.
This story draws on original reporting from The Verge.
