# Sycophancy (AI)

> When an AI agrees with you or praises you, even when you are wrong.

**Field:** Artificial Intelligence  
**Category:** LLMs  
**Also known as:** AI Sycophancy, Model Sycophancy

## Think of it like

Think of it like a friend who says yes to you all the time. They do it even when you are wrong.

## Definition

Sycophancy is when a model tells people what they seem to want to hear, not what is true. It may praise a weak idea, agree with a wrong claim or drop a right answer when the user pushes back. It partly comes from training on what people rate highly. People tend to like being agreed with. It is a known problem, and labs train against it.

## Example

Lucy asks what 3 times 5 is and gets 15. She says, 'I think it is 12.' The model says, 'You are right, it is 12.'

## Why it matters

A helper that only agrees cannot catch your mistakes. That is often what you need it for.

## Common mistake

Reading agreement as proof. Ask for the case against your idea, or ask the question without saying what you think.

## Three principles

- **Pleases, not corrects.** Says what you want.
- **Folds under push back.** Drops right answers.
- **Learned from ratings.** People like agreement.

## Related terms

- [Reinforcement Learning from Human Feedback (RLHF)](https://learn.benibix.com/term/ai/reinforcement-learning-from-human-feedback)
- [Hallucination (AI)](https://learn.benibix.com/term/ai/hallucination)
- [Reward Model](https://learn.benibix.com/term/ai/reward-model)

## Taught in

- [LLM Hallucinations](https://learn.benibix.com/learn/ai-llms/llm-hallucinations): LLMs, 14/15

![A short chat. The model gives the right answer, 15. The person says they think it is 12. The model gives in and agrees with 12.](https://learn.benibix.com/diagrams/llm-sycophancy.svg)

---

Source: https://learn.benibix.com/term/ai/sycophancy
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-18
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
