Anthropic’s Claude AI appears to be facing another wave of issues, with users reporting failed requests and “529 Overloaded” messages when trying to use the service on Monday morning. The error indicates a temporary inability to service requests and directs users to Anthropic’s status page for more information.
Increased errors in many services
Anthropic itself confirmed that it was experiencing a broader technical incident. At 05:06 UTC on August 24, the company announced that it was investigating increased error rates affecting requests for the Claude Mythos 5, Claude Fable 5, Claude Opus 5 , and Claude Opus 4.8 models.
See also: Claude: Voice chat feature now available on Opus and Sonnet models
The problem, according to the relevant update, was not limited to the chat app . claude.ai, Claude API, Claude Code, and Claude Cowork were also affected .
Shortly afterwards, at 05:27 UTC, Anthropic said it had identified the cause of the problem and was working to restore services. At 06:42 UTC, the company noted that restoration work was ongoing. In its latest update, it said the problem had been resolved.
The "529" wasn't necessarily the cause
Of particular interest is the distinction between the message that users saw and Anthropic's official description. "529 Overloaded" was the indication that some users received, but the company has not confirmed that overload was the actual cause of the overall problem.
In other words, the error describes the result the user is seeing, not necessarily the technical problem behind it. Until more information, any connection to a lack of computing power, increased demand, or a specific infrastructure malfunction would be just speculation.
A difficult August for Anthropic
The new incident adds to a series of issues that have been reported in August. On August 18, degraded performance affecting multiple models, while on August 19, a separate incident involving Opus 5 and Haiku 4.5. Another outage, affecting multiple models, was reported on August 20.
See also: AWS Continuum: Secure code across Claude Code, Codex, and Kiro with a unifying AI orchestrator
Meanwhile, users in online communities have reported access issues and recurring “529” errors. Such reports don’t necessarily reveal how many users are affected, but they do show how quickly a malfunction in an AI service used every day can be detected.
When a problem affects the entire ecosystem
Claude's case highlights a broader issue for modern AI platforms. Claude is no longer just a chatbot. It is an entire ecosystem of services, from simple chat and API to programming tools and professional applications.
When a problem occurs in a shared infrastructure, the consequences can be transmitted simultaneously to different products. For a simple user, this may mean a failed conversation. For a developer, however, it may mean failed API calls, interruption of automated processes , or delays in an active project.
Reliability becomes a critical issue
The incident is a reminder that as AI becomes more integrated into everyday work, availability becomes more important. Businesses and developers are no longer just evaluating the quality of a model, but also the stability, speed, and resilience of the platform that hosts it.
