Claude down: users report failures in chat, code and Anthropic app
Users of Claude AI, Anthropic’s artificial intelligence assistant, face access problems this Tuesday (23). Reports focus on failures in Claude Code, the main chat and the application, according to data from monitoring platforms.
Downdetector records spikes in complaints in the last few hours, with Claude Code representing around 46% of complaints, followed by Claude Chat (27%) and the app (15%). Many users report blank responses, extreme slowness, or inability to start conversations.
More on this story: Anthropic identifies problem and implements correction in Claude chatbot
High demand puts pressure on infrastructure
Anthropic, the company behind Claude — which includes models such as Opus, Sonnet and Haiku — has been dealing with accelerated user growth. In recent months, the platform has recorded multiple outages, many linked to the difficulty of scaling the infrastructure to meet growing demand, especially among developers and companies that use Claude Code for programming.
This instability is not isolated. In June, the service already accumulated several partial outages, which raises questions about Anthropic’s ability to maintain stability while competing with rivals like OpenAI. The API for developers and companies also appears in some reports, although it is not always the most affected.
See Also
New official image of the Galaxy Tab S12 Ultra exposes S Pen design and features
Acclaimed game, The Witcher 3: Wild Hunt will leave Xbox Game Pass on August 31st
New Honda CB1000 Hornet 2027 arrives in Brazil in September
Full coverage: Latest News (EN)
What changes for those who use it
For professionals who depend on Claude on a daily basis — whether for coding, data analysis or automation —, outages interrupt workflows and force temporary migration to alternatives. Users on X (formerly Twitter) mention frustration when seeing the service go down precisely at times of high productivity.
Anthropic has not yet released an official statement regarding the current incident, but its status page tends to update quickly when problems are detected. In previous outages, the company usually attributes part of the failures to the unprecedented volume of accesses.
