AI chatbots can be guilt-tripped into granting requests, study finds

Original: Door-in-the-Face Requests and Refusal Behaviour in Large Language Models

Source: arxiv.org ↗