Three frontier AI services fail together

Three major AI assistants—ChatGPT, Claude, and Grok—experienced outages in the same morning on September 3, with OpenAI, Anthropic, and xAI each reporting service problems. The incident timeline records ChatGPT errors around 11 a.m. ET, Claude’s partial outage resolving around 12:15 p.m., and Grok failures beginning around 9:30 a.m.; all three later recovered. Teams running agents should treat provider failover, resumable jobs, and explicit model portability as production requirements, not disaster-recovery polish, as the archive’s portability analysis argues.