ChatGPT and Sora suffer 10‑hour global outage

• Global Disruption Across All Tiers:

  The outage began around 2:45 a.m. ET (9:45 a.m. SAST), affecting all users—free, Plus, and API—along with the Sora video model. Users across North America, Europe, Asia, and Australia reported errors such as “too many requests,” “error in message stream,” and complete unresponsiveness. Notably, only OpenAI’s Playground tool remained fully functional during the first hours.

• Downdetector Spikes Indicate Widespread Impact:

  Downdetector showed more than 1,900 incident reports at the peak, with 93% tied directly to ChatGPT failures. A second spike followed as morning hours hit more global regions. The error reports reflected the cascading impact across time zones.

• Delayed and Fragmented Communication from OpenAI:

  OpenAI did not immediately acknowledge the issue on its X (Twitter) account, despite publishing updates on its status page, confirming “elevated error rates and latency” by 3 a.m. ET. A mitigation was announced by 9:43 a.m. ET, with vague references to internal causes—likely a mix of server overload, rate limiting, or instability from feature rollouts.

• Repeat Outages Undermine Reliability:

  This incident follows several previous outages tied to platform expansions:

  • March 2024: Disruption after GPT-4o access was extended to free users

  • December 2024: Blackout during integration with Apple Intelligence

  • November 2023: A major DDoS attack linked to suspected Russian actors

    These cumulative events cast doubt on OpenAI’s infrastructure readiness for scale and resilience.

• Sora, API Platform, and GPT Models Affected:

  Status indicators showed:

  • ChatGPT: 21 components failed

  • API platform: 14 affected

  • Sora: 4 modules offline

    GPT-4o was widely unusable, while o3 and 4o-mini were intermittently accessible and suggested as workarounds.

• Enterprise & Educational Use Disrupted:

  Productivity across education, journalism, development, and marketing was heavily impacted. Many students in the U.S. and Canada—granted free ChatGPT Plus in April—found themselves without key learning tools. Professionals relying on GPT workflows for research and summarization lost valuable time and context.

• Social and Cultural Reaction:

  The absence of ChatGPT triggered a flood of memes, humor, and mild existentialism on social media.

  • One viral joke dubbed it “2018 Mode: Activated.”
  • Users staged mock funerals, panic-posted alternatives, and questioned their AI dependence.
  • Outage discussions trended on X and Reddit globally.

• Competitor Tools Gained Traction:

  Google Trends (via QR Code Generator PRO) revealed:

  • DeepSeek: 109% spike in searches (2.13 million)

  • Claude AI: 95% jump (291,181 queries)

  • Gemini and Grok also saw spikes as users tested them for stability and coverage.

    The sudden redirection in user traffic highlighted the fragility of platform loyalty under duress.

• Outage Duration and Partial Recovery:

  The outage extended beyond 12 hours. ChatGPT status transitioned from red to yellow by late afternoon, signaling partial restoration. Despite core functionality returning, some features like Read Aloud continued experiencing errors. OpenAI’s internal systems remained under monitoring throughout the night.

Why This Matters:

This extended global outage reveals critical weak points in the operational backbone of today’s most widely used AI platform. With over 500 million users integrating AI into daily workflows, interruptions ripple across education, enterprise, and personal productivity. Repeated system failures raise concerns not only about reliability but also about OpenAI’s transparency, contingency planning, and user communications. In an increasingly AI-dependent world, this event reinforces the need for distributed tooling strategies, alternative options, and robust expectation management for future incidents.