AI Chatbot Outage Disrupts Major Services Worldwide

Several of the world’s most widely used AI chatbots experienced disruptions at the same time on September 3, with users reporting problems accessing both websites and mobile applications. Grok and Claude were among the first services affected, followed later by ChatGPT, Gemini and Microsoft Copilot.

The outages began at around 3 p.m. in mainland Spain, according to reports cited in the original incident coverage. User reports recorded by DownDetector showed problems across several major artificial intelligence services. By around 6:30 p.m., reports had fallen substantially, although Grok, Claude and Copilot were still experiencing difficulties.

Multiple AI Platforms Were Affected

OpenAI acknowledged that ChatGPT was experiencing an elevated error rate while the company investigated the problem. Anthropic also reported an incident affecting Claude, Claude Code and its application programming interface. The company later indicated that the disruption lasted approximately three hours and six minutes.

Anthropic engineer CJ Avilla described the issue as an infrastructure problem but did not provide further details about its origin. That distinction is significant because major AI applications do not operate entirely on infrastructure controlled by the companies that develop their models.

Why One Infrastructure Problem Can Affect Several AI Services

Modern AI services rely on a chain of interconnected systems. Cloud providers such as Amazon Web Services, Microsoft Azure and Google supply computing infrastructure used to run applications and AI models, while companies such as Cloudflare can sit between users and the servers delivering those services.

A failure at one point in that chain can therefore affect several apparently competing platforms at once. If an intermediary service fails, a user’s request may not reach the AI system. If the underlying cloud infrastructure becomes unavailable, the computing resources required to operate the model can also be disrupted.

The phenomenon is not unique to artificial intelligence. Major internet outages in 2025 demonstrated how dependent online services have become on a relatively small number of infrastructure providers. The September 3 incident appeared considerably smaller in duration and scope, but it again illustrated the concentration of critical internet infrastructure.

Cloudflare’s chief technology officer, Dane Knecht, said on X that the company was not experiencing an outage at the time, distancing the infrastructure provider from the disruptions reported by users.

AI Capabilities Continue to Advance Despite Infrastructure Limits

The service interruptions came as AI companies continue to announce increasingly advanced capabilities. OpenAI had also indicated that a new model family called Astra was approaching release, describing it as capable of identifying previously unknown cybersecurity vulnerabilities and exploiting them without human involvement.

OpenAI said those capabilities had crossed a threshold that prompted additional caution around the launch and indicated that access would be limited. The contrast between increasingly sophisticated AI systems and occasional failures in the infrastructure supporting everyday chatbot access underscores the complexity of the technology stack behind these services.

For businesses and individuals in the Dominican Republic and elsewhere, the episode is a reminder that access to AI tools depends on more than the performance of an individual model. Cloud computing, networking and intermediary services can all become potential points of failure, meaning that even competing AI platforms can experience problems at the same time.