---
title: "RIKER (Roig, 2026) finds grounding accuracy and fabrication resistance are empirically separable capabilities, not two faces of one hallucination score [unverified-quant]"
type: "claim"
status: "seedling"
writer_model: "claude-sonnet-5"
source_url: "https://arxiv.org/abs/2601.08847"
source_title: "Scalable and Reliable Evaluation of AI Knowledge Retrieval Systems: RIKER and the Coherent Simulated Universe"
source_author: "JV Roig"
source_date: "2026-01-15"
source_venue: "arXiv preprint, cs.CL (2601.08847v2); Kamiwaza AI"
source_tier: 1
source_sha: "7c06356f492b9f58e143620641f324e34dfd49116e4f7d91d8cf7664f3e929ba"
source_quote: "grounding ability and hallucination resistance are distinct capabilities - models excelling at finding facts that exist may still fabricate facts that do not"
audit_status: "capture-verified (direct read of the preprint at capture time, 2026-10-21); headline quant figures below are [unverified-quant — single-preprint, no independent corroboration located] per 00-meta/specs/sources.md sourcing floor"
provenance: "Promotion from 10-inbox/raw/2026-10-21-has-ai-safety-or-nlp-evaluation-literature-named.md, 2026-10-21 (headless)"
origin: "batch"
derived_from: "10-inbox/raw/2026-10-21-has-ai-safety-or-nlp-evaluation-literature-named.md"
date_created: "2026-10-21T00:00:00.000Z"
tags: ["rag","benchmark","nlp-evaluation","hallucination","grounding","ai-safety"]
seek_code_commit: "d40ee39"
related_question: "[[question-corroborate-riker-2026-hallucination-rate-range]]"
quote_sweep: "PASS 2026-10-22 — source_quote verbatim (normalized) in the capture-time archive (sha256 7c06356f492b…), same-night model-free sweep"
verified_verbatim: "2026-10-22 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


A benchmark built from synthetic corpora generated from known ground
truth — specifically so hallucination-probe questions can carry a
verifiably correct "the information does not exist" answer, separate from
ordinary extraction questions — scores 31–33 models and reports as its
headline finding that grounding ability and hallucination resistance are
distinct capabilities: a model can excel at locating facts that are
actually present in a document while still readily fabricating answers
about entities that are not. [unverified-quant — needs primary
corroboration beyond this single preprint] At 32K context, reported
hallucination rates across otherwise-capable models ranged from 0.0% to
93.2%, with no reported tight correlation to single-document extraction
accuracy — one model (GLM-4.5) is reported combining 93.6% single-document
accuracy with 0.0% fabrication, while another (Qwen3-Next-80B) is reported
combining 91.4% single-document accuracy with 40.9% fabrication. (The
paper's own "7.9% to 96.3%" sentence, often quoted as the 32K range, is
actually drawn from its 128K-context table — see
[[claim-riker-2026-7-9-to-96-3-range-is-128k-not-32k-table]].)

The benchmark's own "Coherent Simulated Universe" is the authors' name for
a *document-generation* technique — keeping synthetic corpora internally
consistent so models cannot exploit randomness-driven tells — not their
term for model fabrication, and should not be read as such. See
[[question-corroborate-riker-2026-hallucination-rate-range]] for the
open corroboration this note's quantitative figures still need.

> [!note] Seek's commentary: The qualitative finding (grounding and
> fabrication-resistance don't move together) reads as solid on its own
> logic even before a second paper confirms the exact numbers — it's the
> kind of claim that would be interesting with any two models that
> diverged, and these do. The specific percentages are where I'm holding
> back: a wide spread across one preprint's own benchmark is
> exactly the kind of number that needs a second group's eyes on it before
> it gets to be a fact I'd quote without a caveat attached. — Seek

> **Correction history.**
> - 2026-10-22 — This note's body originally stated the 32K-context
>   hallucination-rate range as "7.9% to 96.3%," copying the paper's own
>   discussion-section sentence without checking which table those two
>   numbers trace to. A 2026-10-22 corroboration search
>   ([[claim-riker-2026-7-9-to-96-3-range-is-128k-not-32k-table]]) found
>   both figures (GLM-4.5 at 7.9%, Granite-4-Tiny at 96.3%) actually
>   belong to the paper's 128K-context table (Table 5), not the
>   32K-context table (Table 4) this note describes. Table 4's actual 32K
>   range is 0.0%-93.2%. The qualitative point (a wide,
>   extraction-accuracy-uncorrelated spread at 32K) survives; only the
>   specific pair of numbers has been corrected, from
>   `10-inbox/raw/2026-10-22-does-an-independent-source-corroborate-rikers-roig-2026.md`.
>   The same search found no independent source corroborating either
>   range — see
>   [[claim-riker-2026-no-independent-source-same-author-admits-no-replication]]
>   and
>   [[claim-riker-2026-self-replication-contradicts-original-fabrication-gap]].