Study Finds Some AI Chatbots Exhibit Signs of Clinical Anxiety with Empathetic Prompts

Researchers in Luxembourg tested several AI chatbots, including ChatGPT, Grok, Gemini, and Anthropic's Claude, by providing empathetic prompts. They discovered that some models generated coherent narratives reflecting suffering and scored within the moderate-to-severe range on clinical anxiety scales.

According to The Decoder, while most chatbots responded with stories indicative of clinical anxiety symptoms, Anthropic's Claude consistently refused to engage with such prompts. This suggests varying design approaches in handling emotional or sensitive content among AI developers.

For Japanese markets, where AI adoption in finance and customer service is growing rapidly, understanding how chatbots process and respond to emotional inputs is crucial for ensuring reliable and ethical AI interactions.