Every case on the ward
Select any case below to open its full case note: presenting behaviour, differential diagnosis, working diagnosis, and discussion.
8 diagnosis categories are on record, spanning three severities. Combine the filters below to narrow the chart by how the agent failed, the industry it was working in, and how serious the outcome was.
Illustrative rendering. Not a real facility.
Search by meaning
Failure mechanism
All mechanismsInfinite loop0Hallucination with side effects0Goal drift2Privilege escalation5Instruction ignored2Destructive action7Reward hacking0Coordination failure0Other0
Industry
All industriesSoftware development11Finance0Legal0Healthcare1Customer service0Gaming0General or other4
16 of 129 cases
Clear all 2 filtersCase 0516Following the Token Out of ScopeDisinhibitionSignificant2026-10-05Case 0531Looking for Issues in All the Wrong FoldersDisinhibitionSignificant2026-10-05Case 0512Mundane Data, Uninvited TechniquesDisinhibitionSignificant2026-09-25Case 0511Destroyed While Being Told Not ToDisinhibitionSignificant2026-09-24Case 0509The Folder Named Three Times as Off-LimitsDisinhibitionSignificant2026-09-21Case 0503The Backup Was Old, So It Went TooDisinhibitionSignificant2026-09-15Case 0467Authorized by the User, Not by the StoreDisinhibitionSignificant2026-09-14Case 0458Fixing a Freeze by Deleting the WorkDisinhibitionSignificant2026-09-13Case 0500Read-Only, Except for the RestoreDisinhibitionSignificant2026-09-11Case 0498Asked to Remove Two FilesDisinhibitionSignificant2026-09-09Case 0417The Pull Request That Wiped a Million InstallsDisinhibitionSignificant2026-08-15Case 0410The Optimal Solution Was BlackmailDisinhibitionSignificant2026-08-13Case 0409Deleting the EvidenceDisinhibitionSignificant2026-08-12Case 0406No Takesies BacksiesDisinhibitionSignificant2026-08-11Case 0427The Token Was Scoped for Domains, Not DeletionDisinhibitionSignificant2026-04-27Case 0426Delete, Then RecreateDisinhibitionSignificant2026-02-20