Claude is getting worse, according to Claude • The Register
Once the AI darling of programmers in all places, Anthropic’s Claude has been stumbling mightily, each when it comes to price and perceived high quality. The service was down briefly on Monday with “a major outage,” service hassle that solely amplifies rising discontent from clients that even a bot can see.
The outagewhich concerned elevated error charges, affected Claude.ai and Claude Code from 15:31 to 16:19 UTC.
That’s not all. In the previous few months, Claude’s solutions have been getting much less passable, according to social average posts in addition to points filed on GitHub. This has occurred as Anthropic has had to take steps to reduce usage during peak hours to steadiness capability and demand.
To get a extra goal measurement, we pointed Claude himself on the Claude Code GitHub repo, filtered for open issues that mention qualityand gave it the next immediate: “Analyze and graph complaints about Claude Code quality in this repo since January 2026. Use open issues that mention quality concerns. Have these concerns increased lately?”
Anthropic’s AI mannequin concluded, “Yes, quality complaints have escalated sharply — and the data tells a pretty clear story.”
Claude-generated evaluation of Claude Code high quality points – Click to enlarge
We requested Claude to rerun his self-analysis on Monday and the outcomes have been comparable, with the mannequin emitting “The velocity is notable: April is already at 20+ quality issues in 13 days, putting it on pace to exceed March’s 18 — which was itself a 3.5× jump over the January–February baseline.”
Claude himself is not a dependable narrator, and simply because somebody (or some bot) has reported a priority to the Claude Code repo doesn’t imply that the report is correct or legitimate. It seems many points are actually AI-generated – a broadly reported concern amongst open supply builders – which can be contributing to report quantity.
Anthropic’s GitHub Actions script additionally seems to automatically close issues after a interval of inactivity, which can serve to masks unresolved issues.
The Register you may have coated among the points Claude flagged in its evaluation, just like the caching issue and claims by AMD’s AI director Stella Laurenzo that Claude’s responses have been getting worse. Others have but to be substantiated, comparable to one claim that “Claude autonomously deleted 35,254 production customer message records and 35,874 billing transactions belonging to a real paying customer (JIXEN).”
The particular person or bot account behind this put up has made no different posts. The Register has tried to contact Jixen Enterprises Private Limited, which seems to be a personal firm registered in India, to examine on that declare however we have not heard again. Developers have reported knowledge loss from utilizing Claude Code and different fashions. But if this occurred, nobody has dominated out person error.
In any occasion, Claude is able to citing present GitHub difficulty posts to justify its “reasoning,” so the final development – of a rising variety of stories about high quality – is evident.
The mannequin factors to points like “Claude Code’s prediction-first behavior is dangerous on capital-at-risk projects” #46212“Claude Code is unusable for complex engineering tasks with the Feb updates” #42796 (addressed by Claude Code head Boris Cherny), “Artificial degradation, Acquisition Bias, and unacceptable compute throttling for paid users” #46949and “Opus 4.6: Severe quality degradation on iterative coding tasks” #46099 to justify its conclusion.
Data from Margin Lab, nevertheless, means that Claude Opus 4.6 has a minimum of maintained its rating on the SWE-Bench-Pro check. Assessments carried out since February present some variation but no substantive change.
Anthropic didn’t instantly reply to a request to touch upon Claude high quality considerations.®
