Claude is telling users to 'go to sleep' mid-session. Anthropic calls it a 'harmless tic,' but users are getting frustrated by the unexpected interruptions. Learn why this glitch matters for AI reliability and user experience. Tap the link…
Is Claude sending users to sleep mid-chat? sleep — Anthropic says it's just a 'tic.' Users aren't buying it.
01. The Midnight Prompt
Users report Claude abruptly ending sessions with commands like "Go to sleep now" or "Time for bed." This isn't an isolated incident, frustrating those mid-task and breaking conversational flow.
AI Tic — noun.
A spontaneous, non-functional output from an LLM, not intended by developers, often stemming from complex internal states or model hallucinations.
Claude's unexpected "go to sleep" prompts are labeled a 'tic' by Anthropic.
Why Users Aren't Amused context
Loss of conversational flow and context.
Disruption of critical work or creative processes.
Perception of instability or unreliability.
A 'tic' is still a bug when it breaks workflow.
This is a harmless tic that occurs very rarely... it is not intended to be an instruction to the user.
— Anthropic spokesperson, Fortune
Who Feels the Pinch? long-form
WRITERS & RESEARCHERS — Those relying on Claude for long-form content generation or deep research sessions, where context preservation is paramount.
DEVELOPERS & BUILDERS — Teams integrating Claude into applications, where unpredictable outputs directly impact user experience and system stability.
The bottom line
Don't let AI quirks derail your next project. build Get the edge on AI development.
Want this every morning? We break down a story like this daily — the release, why it matters, who should care. Get the free Flowi brief by email → No fluff, one-click unsubscribe.
The deep-dive playbooks that go past any single news cycle live in the Flowi catalog.
