SAN FRANCISCO, Sept. 3 (Xinhua) -- ChatGPT, Claude and Grok were among several major U.S. artificial intelligence (AI) services hit by near-simultaneous outages on Thursday, disrupting access for some users before affected services recovered over the next several hours.
Anthropic's official status page showed elevated request error rates across multiple Claude models during the morning. The disruption affected Claude's website, API, Claude Code and Claude Cowork. The company later said the outage was caused by an infrastructure issue and that the impact had ended by 9:16 a.m. Pacific Time (1616 GMT).
Grok, developed by SpaceXAI, also suffered disruptions around the same period. The company's status page showed that its U.S. East and West APIs, web and mobile services, as well as Grok on social media platform X, were affected, with some services disrupted for about three and a half hours.
SpaceXAI later confirmed that an outage occurred at its compute center in Memphis, Tennessee, and apologized to its "impacted compute partners." The company said all systems were subsequently restored.
OpenAI also reported elevated error rates affecting ChatGPT and its AI coding tool Codex. A routing error beginning at around 7:43 a.m. Pacific Time (1443 GMT) prevented some users from accessing ChatGPT and Codex across multiple platforms, and a fix was implemented at around 8:17 a.m. (1517 GMT), the company said.
OpenAI's status page later recorded another period of elevated error rates affecting ChatGPT and Codex. The company said the incident had been resolved after mitigation measures.
The outages at the three major AI providers overlapped in time. While SpaceXAI said some of its compute partners were impacted by the Memphis outage, publicly available information has not confirmed whether the outages at OpenAI, Anthropic and SpaceXAI shared a common cause. ■