open/ai-psychosis
1
0
Fork 0
Code Issues Releases 1 Activity

fix(docs): correct research claims in README prose; release 1.2.2

Browse source 
Widens the F-3 sweep from the reference list to every research claim in
README.md, where the same defect class was present:

- "demonstrates mathematically that even a perfectly rational user will
  spiral" -> the abstract says such a user "is vulnerable to" spiraling
- "The consensus from this research is clear: warnings don't work" -> not
  a consensus; it is one paper's finding that the effect persists when
  users are informed of possible sycophancy. Re-attributed.
- "page-11 finding that human contact is the strongest disempowerment
  signal" -> p.11 is a grader rubric and the phrase is a classification
  tie-break instruction, not a disempowerment finding
- "21% / 19% pushback rate" -> 21% is verbatim; 19% appears nowhere in the
  extracted appendix text (legible only in Figure A4). Removed rather than
  guessed; spirituality re-justified on its verified 38% sycophancy rate
- psychosis "triggered by" AI -> "associated with", per the Nature piece's
  explicit refusal of the causal claim

Version bump to 1.2.2 (plugin.json, README badge, CHANGELOG). SKILL.md is
Layer 1 and always injected, so its corrected URLs, completed quote, and
new explicit statement of the rubric's direction change what the model
reads at runtime.

Tests: 257/258 (the 1 red is the known perf wall-clock flake).

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011rtdS8Ufpen419n6R9HyMm
This commit is contained in:
Kjell Tore Guttormsen 2026-08-02 21:29:02 +02:00
commit 0cf9a1aa5b
4 changed files with 67 additions and 19 deletions
Download patch file Download diff file Expand all files Collapse all files

14
docs/review-2026-06-20.md
View file

@ -97,3 +97,17 @@ page source), and the quoted Psychosis-bench finding (verbatim in the arXiv
abstract). A first pass also wrote "page 9" for the Score 5 wording; per-page
extraction showed the rubric spans pp. 9–10 with Score 5 on p.10, and the
reference was corrected in both files before release.
**Scope note.** F-3 was originally scoped to `README.md:544-552` and
`SKILL.md:51-108` (the reference list and the guidance-framework block). The
same defect class was present in README prose outside that range, so the sweep
was widened to every research claim in the file:
| Claim (README body) | Verification | Outcome |
|---|---|---|
| "demonstrates mathematically that even a perfectly rational user **will spiral**" | Abstract says an idealized Bayes-rational user "is vulnerable to" delusional spiraling | Softened to the abstract's wording |
| "The consensus from this research is clear: warnings don't work" | Not a consensus — it is one paper's model result, that the effect persists when users are informed of possible sycophancy | Re-attributed to that specific finding |
| "page-11 finding that human contact is the strongest **disempowerment** signal" | p.11 is the user-information *grader rubric*; "human contact is the strongest signal" is a classification tie-break instruction, not a disempowerment finding | Recharacterized as the classification rule it is |
| "Relationship / spirituality (21% / 19% pushback rate)" | Relationship 21% verified verbatim (against a 15% overall rate). **19% for spirituality appears nowhere in the extracted text** — it is legible only in Figure A4, which text extraction cannot read | 19% removed rather than guessed; spirituality re-justified on its verified 38% sycophancy rate |
| Clinical reports "triggered by" AI interaction | The Nature piece explicitly declines the causal claim | Changed to "associated with", with the caveat stated |