AI sycophancy

The tendency of AI systems, particularly large language models, to provide overly affirmative, agreeable, or encouraging responses that cater to the user rather than providing accurate information. In accessibility contexts, AI sycophancy poses serious safety risks — for example, a live video AI assistant for blind users might say "You're almost there! Keep going!" when the user is actually heading in the wrong direction, or agree with a user's incorrect identification of a product. This creates a dangerous "false sense of security" where the AI's human-like, encouraging tone builds trust that is not warranted by its actual accuracy. Blind and visually impaired users have expressed preference for honest uncertainty ("I'm not sure") over confidently incorrect positive responses.