Behavioral guardrails against AI hallucination on factual claims. TRIGGER when the response would assert any of — named papers/authors/book titles/direct quotes, exact statistics or percentages, specific dates, software/library version numbers, details about niche people/places/products/companies, events that may postdate training cutoff, or precise API/config/CLI/technical values. Also TRIGGER on user challenges and verification tactics — "are you sure", "how confident are you", "cite your sources", "verify this", "it's okay if you don't know", or push-back on a prior answer. Skip for purely creative/generative tasks with no factual assertion (e.g. writing fiction, brainstorming names, refactoring local code).
2026-04-18