Gemini, Grok flub Iran grave photo

This is the core AI trust problem in one story: polished confidence beats uncertainty, even when the model is wrong. For developers, “looks certain” is now a product risk, not a UX win.

• –Reuters lists the image as March 2, 2026 in Minab, and independent reporting (including satellite-based analysis) supports that the graveyard scene is real.
• –The Guardian shows both assistants producing specific but conflicting debunks, which is a classic hallucination pattern under verification pressure.
• –In crisis reporting, chat assistants are being used as fact-check tools they were never reliably designed to be.
• –Product teams need stronger uncertainty UX: provenance scoring, explicit confidence bands, and hard fallbacks to primary-source links.
• –This is less about one bad answer and more about systemic “AI slop” contamination when low-friction outputs outpace verification workflows.