Claude AI Suffers Major Outage on Wednesday, Thousands of Users Without Access
Status page confirmed elevated errors on all models, investigation ongoing
Status page confirmed elevated errors on all models, investigation ongoing
Chatbot Claude, one of the most acclaimed AI assistants, suffered a major outage on Wednesday, July 29, 2026. According to information reported by Hacker News, Claude's status page (status.claude.com) announced that it is recording "elevated errors on all models", and "We are currently investigating this issue" was stated in their report at 19:49 UTC. The problem affects all key services: the claude.ai interface, the API, the Claude Code tool for developers, and the new Claude Cowork platform.
On the independent website DownDetector, which tracks internet service outages, nearly 4,000 users reported difficulties, meaning thousands of people were unable to access the tool, reports CNET. When trying to access, CNET journalists received a message: "Due to unexpected capacity constraints, Claude cannot respond to your message. Please try again soon."
CNET reports that a representative of San Francisco-based Anthropic did not immediately respond to a request for comment. Insider Paper also briefly posted on its Telegram channel at 22:04 UTC: "BREAKING: Claude is down". CNET's article about the outage was published at 22:45 UTC the same day.
The status page also recorded errors that occurred on Monday, July 27, 2026, on the Claude Opus 5 (the powerful flagship of the series) and Haiku 4.5 (a faster, lighter model). Those errors were resolved the same day, and Tuesday, July 28, passed without incident.
CNET once declared Claude the best AI chatbot, praising its consistency and noting that it is often considered more business-oriented than a general assistant. Anthropic was founded in 2021 by former OpenAI employees, the company behind the popular ChatGPT.
The outage from Wednesday evening is still under investigation. Croatian users who rely on Claude daily are also affected, partly because all models, including those for advanced tasks such as coding and real-time collaboration, are out of service until the issue is resolved.