Incident Details:

  • Affected Regions: Reports of outages came from multiple countries, including the US, Canada, Denmark, the UK, and Russia.
  • Timing: The issue began around 3:00 AM Moscow time on December 27.
  • OpenAI’s Response: The company acknowledged the problem on its status page, citing "a high rate of errors" across its systems. Engineers worked to identify and resolve the issue, restoring services within a few hours.

Potential Cause:
While OpenAI did not name the "upstream provider" involved in the incident, speculation centers on Microsoft, OpenAI's exclusive cloud service provider. Coincidentally, Microsoft reported a "power issue" in one of its data centers affecting North America at the same time as OpenAI’s outage. Xbox Cloud Gaming services also experienced disruptions during this period. Microsoft resolved the power issue approximately three hours later, stating that the affected data center had been "fully restored."

The outage highlights the vulnerabilities of AI services reliant on cloud infrastructure. While OpenAI quickly restored functionality, the incident underscores the critical role of robust cloud service management in ensuring the reliability of AI-driven platforms like ChatGPT and Sora.