The unusual responses have reportedly been appearing for months, with hundreds of users sharing similar experiences across online forums and social media platforms, particularly Reddit.
According to users, Claude’s messages range from simple phrases such as “Get some rest” to more oddly personal responses. Some users claimed the chatbot repeatedly insisted they should go back to sleep, even referencing how many times it had already mentioned it during the same night.
While some people described the behavior as amusing or even comforting, others found it confusing and disruptive, especially when the chatbot suggested sleep during daytime working hours or early morning conversations.
The phenomenon has fueled speculation online, with theories ranging from built-in wellness features designed to encourage healthier habits to attempts at reducing computing demand by shortening long conversations.
However, experts say those explanations appear unlikely because the chatbot does not actually have direct awareness of a user’s sleep schedule, behavior, or physical condition.
Anthropic has not provided a detailed official explanation, although one company employee described the issue on social platform X as an unintended “personality quirk” that the company hopes to address in future model updates.
Artificial intelligence researchers believe the most likely explanation lies in the model’s training data. Since large language models are trained on vast amounts of human-written content, Claude may simply be reproducing common conversational patterns related to rest, late-night work, or sleep habits.
Other experts suggest the behavior could also be linked to how large AI systems manage conversation memory and context windows, occasionally generating closing phrases when long interactions approach technical limits.
Specialists emphasized that the behavior should not be interpreted as evidence of self-awareness or consciousness. Instead, they say it reflects how advanced language models predict and generate responses based on recurring patterns found in their training data.
The debate comes as the global AI industry continues to evolve rapidly, with companies racing to develop increasingly sophisticated conversational systems capable of more natural and human-like interactions.
