OpenAI、Anthropic和xAI同时遭遇服务中断，原因不明

美国东部时间周四上午，OpenAI、Anthropic和xAI三家AI公司的主流模型几乎同时发生服务中断1。ChatGPT、Claude和Grok等聊天机器人无法使用，影响范围广泛1。

具体而言，OpenAI在美国太平洋时间上午7:43分因路由错误导致ChatGPT和Codex部分用户受影响，于8:17分解决1。Anthropic的中断始于美国太平洋时间上午6:23分，涉及Claude Mythos 5.1、Claude Fable 5.1和Claude Opus 5等多个模型版本，于上午9:16分标记为已解决1。xAI旗下的Grok在美国太平洋时间上午6:30分全平台无法访问，直至上午10:05分才恢复1。

对于中断原因，各公司态度不同。xAI指出问题源于孟菲斯计算中心的设备故障1，而OpenAI和Anthropic均未对外说明具体原因1。SpaceX随后发表声明表示"也想为受影响的计算合作伙伴道歉"1。这一表态值得注意，因为Anthropic和xAI在5月曾宣布与SpaceX建立计算合作伙伴关系1。OpenAI发言人Kathleen Chaykowski确认问题已解决并表示持续监测中1。主要云基础设施提供商Cloudflare、AWS和Microsoft Azure均未在同日报告中断事件1。

Major artificial intelligence platforms operated by OpenAI, Anthropic, and xAI experienced rare service disruptions on Thursday morning, September 3, 2024, in the United States, leaving ChatGPT, Claude, and Grok inaccessible to users 1. The three companies restored their services within hours, though OpenAI and Anthropic declined to specify external causes for the incidents 1.

The outages began at staggered times on Pacific Time: Anthropic reported partial disruptions affecting Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5 at 6:23 a.m., xAI reported Grok becoming unavailable across all platforms at 6:30 a.m., and OpenAI experienced routing errors preventing ChatGPT and Codex access for some users starting at 7:43 a.m. 1. Resolution came sequentially, with OpenAI resolving its problems by 8:17 a.m., Anthropic marking issues as resolved at 9:16 a.m., and xAI restoring Grok by 10:05 a.m. 1. OpenAI spokesperson Kathleen Chaykowski stated the issue had been addressed and the company was maintaining continued monitoring 1.

xAI attributed its outage to equipment failures at its Memphis computing center 1. SpaceX released a statement apologizing "to affected computing partners" 1. Both Anthropic and xAI announced computing partnerships with SpaceX in May 1. No simultaneous outages were reported from major cloud infrastructure providers including Cloudflare, AWS, or Microsoft Azure on the same day 1.