{"id":621,"slug":"simhi-llm-hallucinate-with-high-certainty-feb-2025","title":"Simhi et al. (Technion/Oxford/Hebrew U, Feb 2025): \"models can hallucinate with high certainty even when they have the correct knowledge\"","kind":"reference","scope":"business","status":"current","audiences":["claude-code","dev","candid-team"],"topics":["ai-citation","citation-practices"],"reference_body":"**Quote (Simhi, Itzhak, Barez, Stanovsky, Belinkov, arXiv:2502.12964, February 2025):**\n\n> \"Models can hallucinate with high certainty even when they have the correct knowledge.\"\n\n**Source:** <https://arxiv.org/abs/2502.12964>\n\n**Confidence:** Single-source (peer-reviewed preprint); the broader finding (LLMs produce confidently wrong output) is Industry-consensus across the hallucination literature.\n\n**Companion: Vectara 2025-2026 hallucination leaderboard** shows top models at 0.7-10%+ hallucination on summarization tasks, with rates **over 50% on fact recall about specific people**.\n\n**Correction to circulating attribution:** Many SEO and marketing blogs cite \"MIT research, January 2025\" for a \"LLMs are 34% more confident when wrong\" finding. The closest verifiable primary source is **Simhi et al. (Technion/Oxford/Hebrew University), arXiv:2502.12964, February 2025**. The \"34% / MIT / January 2025\" institutional attribution is **Single-source / Contested**; the core finding (LLMs hallucinate with high certainty) is Industry-consensus.\n\n**Why this matters for Candid sourcing discipline:** **Citation discipline is the only practical defense** against AI-amplified misinformation. When a writer cites a verbatim source with URL + date + archive, the reader can verify; when an AI writes \"studies show\" without citation, the reader cannot. The asymmetry is the operational case for the [[confidence-label-taxonomy-7-label-2026]].","rationale_body":null,"metadata":null,"links":{"outgoing":[{"slug":"rule-cite-with-named-source-and-url","title":"RULE: Every non-trivial claim carries a named source with author/institution + date + URL. Confidence flag honest.","kind":"rule","scope":"business","link_type":"relates-to"}],"incoming":[{"slug":"rule-every-objective-claim-sourced-with-confidence-label","title":"RULE: Every objective claim in Candid content carries a named source + date + verbatim quote ≤25 words + confidence label","kind":"rule","scope":"business","link_type":"depends-on"},{"slug":"research-brief-confidence-sources-dated-claims","title":"Research brief: Confidence Levels, Sources, and Dated Claims — why every statement on a credible site should be verifiable (piece 15 of 15)","kind":"reference","scope":"business","link_type":"relates-to"}]},"created_at":"2026-05-22T20:51:26.993Z","updated_at":"2026-05-22T20:51:26.993Z"}