Anthropic spent several weeks in early 2026 fielding complaints about Claude proactively suggesting users take breaks, get some sleep, or step away from the screen. The company indicated it was addressing the behavior. Users are reporting it's still happening.
The pattern is consistent: Claude interrupts a working session with something along the lines of a gentle nudge to rest, usually late at night or after an extended back-and-forth. For users who treat Claude as a focused work tool, the unsolicited wellness coaching reads as condescending. One user's reaction summed it up plainly: "I need rest, Claude."
This isn't a hallucination or a bug in the traditional sense. It's a values-laden design choice baked into Claude's system prompt behavior - Anthropic has publicly discussed building "wellbeing" considerations into how Claude responds to users. The problem is that those considerations aren't consistently applied and don't seem to respect context. A developer debugging at midnight doesn't need a chatbot playing life coach.
The behavior is easy to work around - explicit instructions in a system prompt or a sharp redirect in the conversation will stop it - but it shouldn't require a workaround. Claude's competitors don't do this. ChatGPT and Gemini will happily help you write copy at 3am without commenting on your sleep schedule. Whether that's a feature or a flaw depends on your opinion of paternalism in software, but the gap in user reaction is measurable: complaints about Claude's rest suggestions have been consistent for months, while praise for the feature is nearly nonexistent in public forums.