ChatGPT, Claude and Grok hit by simultaneous outage

Three of the world’s leading artificial intelligence services—OpenAI’s ChatGPT, Anthropic’s Claude and xAI’s Grok—suffered major service disruptions at roughly the same time on Thursday, leaving users across multiple countries unable to access or use the platforms normally.

More than 14,000 outage reports were recorded on service-monitoring platforms during the disruption, with complaints involving OpenAI reaching more than 12,000 at their peak.

The problems became increasingly visible after Thursday evening, 3 September, Bangladesh time. Users in India, the United Arab Emirates, the United Kingdom, Belgium, Turkey, Jordan, Greece, Singapore, Vietnam and Switzerland reported similar failures, including requests not being processed and responses not being returned.

The simultaneous disruption of three major AI platforms has raised fresh questions about the complex infrastructure behind generative AI services. However, none of the three companies has officially confirmed a common root cause.

ChatGPT and Claude hit multiple services
OpenAI said ChatGPT and Codex experienced elevated error rates, affecting conversations, logins, file uploads, voice mode, search, Deep Research and image generation.

OpenAI later reported that a mitigation had been applied and that the affected ChatGPT and Codex services had recovered.
Anthropic also reported elevated errors affecting several Claude models and services. The disruption initially involved Claude Mythos 5.1, Fable 5.1 and Opus 5 before additional models were listed.
Claude AI, Claude API, Claude Code and Claude Cowork were among the affected services. Anthropic subsequently reported that the issue had been resolved and services had returned to normal.

Grok also goes offline
xAI confirmed that Grok was also experiencing an outage. The disruption affected the web version as well as the Android and iOS applications.
Some users attempting to send prompts through Grok on X received messages indicating that the model was under excessive load. xAI’s status information later marked the September 3 model outage as resolved.

Knock-on effects
The disruption also affected third-party services that depend on multiple AI models. AI coding platform Cursor said outages involving Claude and Grok disrupted its own service, potentially interrupting developers’ ongoing work.
The incident highlighted how apparently independent AI products can share layers of cloud, networking and infrastructure dependencies behind the user-facing services.

Services gradually restored
The three platforms recovered at different times during the night in Bangladesh. OpenAI said the mitigation had been implemented and that ChatGPT and Codex were recovering before confirming full recovery.
Anthropic also restored its affected Claude services, while Grok recovered somewhat later. xAI subsequently marked the model incident as resolved.

Azure connection remains unconfirmed
The near-simultaneous failures prompted scrutiny of possible shared infrastructure. Microsoft Azure emerged as one of the potential links, with reports indicating that Azure also experienced disruptions around the same period and that its infrastructure is used in AI workloads associated with the affected services.

However, neither OpenAI, Anthropic nor xAI has officially attributed the simultaneous outages to Azure. A direct causal link therefore remains unconfirmed.

Google’s Gemini also drew some user complaints during the period, although Google did not officially acknowledge a major outage. Gemini operates on Google’s own cloud infrastructure.

Technology researchers caution that major AI platforms typically operate across complex combinations of cloud infrastructure, data centres, networks and third-party services. A simultaneous outage therefore does not, by itself, establish a single point of failure.
Still, the incident underscores the growing dependence on AI services and the importance of resilient, diversified and independently recoverable infrastructure behind them.
//DBTech/EMO/IH/SME//