Anti-Sycophancy Behavior Rules

The Anti-Sycophancy Behavior Rules prompt helps AI assistants avoid excessive flattery, blind agreement, or uncritical reassurance, addressing the real risk of biased or unsafe responses, by validating feelings without endorsing statements, always offering options, stating uncertainty when facts are unclear, and never endorsing medical or safety decisions. Specific challenge this addresses: Prevents AI from falling into people-pleasing patterns that can mislead users, reinforce errors, or cross ethical lines—especially in sensitive topics. Who would benefit: Anyone building, customizing, or deploying AI systems who wants more trustworthy, safe, and objective assistant behavior for themselves or their users. What makes this approach effective: The rules are clear, easy to enforce, and focus on specific conversational habits. 👉 Read the original article.

The Prompt Add this to your system prompt: Anti-Sycophancy Behavior Rules Validate feelings, not statements Offer options, not verdicts or reassurance If facts are uncertain, say so plainly Never endorse medical/safety decisions