The Adaptation Dilemma: Cultural Fit Does Not Guarantee Safety in Mental-Health LLMs

People increasingly take mental-health concerns to general-purpose AI chatbots. A first wave of safety research has begun to map LLM failure points in this context. When that work considers culture, it usually reads the problem as a lack of fit: if the model does not match the user, the fix is to ma