SAN FRANCISCO: OpenAI restores ChatGPT services following a multi-hour global outage that disrupted access for over 12,000 users across web and mobile platforms on February 3.
The disruption began around 3:00 PM Eastern Time, affecting conversations, search functionality, image generation, voice mode and API access as elevated error rates cascaded through OpenAI’s infrastructure. The company acknowledged the incident within minutes and deployed mitigations, marking primary issues resolved by 5:14 PM Eastern.
Down Detector recorded complaints spiking from initial reports at 12:08 PM Pacific to peak levels exceeding 25,000 incidents within two hours. Users encountered error messages preventing them from loading conversations, receiving responses or accessing platform services.
OpenAI’s status page confirmed elevated errors for ChatGPT and Platform users while engineers worked to identify root causes. The timing proved particularly disruptive for professionals and developers relying on ChatGPT for daily workflows, underscoring enterprise dependence on continuous AI availability.
The outage follows OpenAI’s Monday launch of ChatGPT Codex for macOS, which CEO Sam Altman reported achieved over 200,000 first-day downloads. Industry observers speculate unexpectedly high Codex adoption may have contributed to infrastructure strain, though OpenAI has not confirmed specific technical causes.
The incident marks ChatGPT’s latest reliability challenge after previous disruptions in January 2025 and subsequent months. Competitor Anthropic experienced separate Claude outages earlier the same day, suggesting broader infrastructure pressures across AI platforms as user bases scale rapidly and enterprises integrate generative models into critical business operations.
Previous
