Always fact-check AI — demand reasons and source URLs

🎯 Why it matters

AI URLs are frequently fabricated — a published 2025 study estimated ~150,000 fake AI citations leaked into scientific literature ("LLM hallucinations in the wild", reported by phys.org / Times of India). Client-facing catch rate is even higher: one hallucination in a deliverable = trust collapse = churn = loss of full client LTV (12-36 months revenue) + cascade to referrals. When AI errs, clients come to you, not Claude.

📊 ROI highlights

• Fact-checking trains you to see AI limits — strengthens your judgment
• 5-year differentiation: who fact-checks wins
• Wrong pivot = 6 months + $50k opportunity cost

⚠️ Common mistakes

• Copy-pasting AI output to deliverable — no verify
• Trusting AI URLs without clicking — 50% fabricated
• "Claude said it, so it must be right" — you are accountable
• Not asking multiple AIs — one's blind spot may be caught by another
• AI fact-checking itself — same prior, useless

✅ Action checklist

• Every key conclusion: ask "Why? Source URL?"
• URLs: click and verify each
• Every key conclusion: cross-check with second AI
• Client deliverable: manual final verify before sending
• Build a fact-check checklist — 5 points where you commonly catch hallucinations