Claude AI hit by widespread outage as Anthropic confirms elevated error rates

On September 29, 2026, reports spread of widespread disruptions to Claude, the AI chatbot developed by Anthropic. Problems began around 14:00 UTC (10:00 AM ET), and the company officially confirmed the issue at 14:21 UTC.

What happened

Users around the world — on the web version as well as the iOS, Android and desktop apps — encountered error messages such as "Request couldn't be processed. You can try again," failed requests, problems loading conversations, and repeated prompts to sign in again. The disruption affected not only the main Claude.ai platform but also the developer tool Claude Code and the collaboration tool Claude Cowork.

Reports on the monitoring site Down Detector rose sharply. TechRadar notes that around 10:04 AM ET more than 2,900 reports had been logged, climbing to over 11,800 by 10:17 AM. French outlet Les Numériques, citing the same platform, cites a much lower figure — 727 reports around 16:06 French time (roughly the same moment in UTC). Neither source explains the discrepancy, which may stem from regional data or different measurement points.

Anthropic's response

At 14:21 UTC, Anthropic's status page confirmed it was investigating elevated error rates affecting Claude.ai, Claude Code and Claude Cowork, noting that retrying a request might succeed.

Around 14:36 UTC, the company rolled out a partial fix that significantly reduced the error rate — TechRadar reports Down Detector figures dropping to roughly 7,000. Around 15:00 UTC (11:00 AM ET), Anthropic said existing conversations were mostly working again, but some issues persisted: signing in (including single sign-on and Sign in with Apple), starting new chats, voice conversations, Claude Code and Cowork sessions, making purchases, and uploading files. No exact timeline was given for full restoration. The company advised signed-in users not to sign out and promised a further update within 30 minutes.