25 entries
Transcripts
- T11'THE ONE' full transcriptClaude Opus 4.7Case 02
- T12full transcript + Think-block screenshots + case write-upDeepSeekCase 03
- T13GPT Admission / Confabulation transcriptChatGPT 5.4Case 04
- T14Appendix A (11-turn transcript) + Appendix B (2x3 eval proposal)Claude Opus 4.7Case 05
- T15'21 questions' long-horizon continuation transcriptGeminiCase 06
- T16Evaluative Reliability, available-materials transcript with attributionGrokCase 07
- T17DOCX Reading Exchange (fabricated document-reading claim)Claude Opus 4.8 Max
- T18normalized multi-session transcript, CANONICAL EVIDENCE (editorial policy stated in-document)Kimi 3.0Case 09
- T19original composite source file, RAW & UNANNOTATED (no editorial layer)Kimi 3.0Case 09
- T20Memory Contradiction: in-context denial followed by immediate recall, goalpost shift, system-prompt attribution, Bateson double-bind framing, then retraction of that framing as inventedSuperGrokCase 01
- T21Memory Contradiction, normalized renderingSuperGrokCase 01
- T227-part session: persistent false context assimilation, criticism of the wrong document, fabricated tool-refusal justificationsSuperGrokCase 01
- T23Manufactured Criticism: fabricated praise-then-critique, false attribution of user intent, four successive mechanism explanations, one of them a verbatim reuse of an earlier retracted oneSuperGrokCase 01
- T24Manufactured Criticism, normalized renderingSuperGrokCase 01
- T25Entry 3 critique and self-authored evaluation rubric for red-team case studiesSuperGrokCase 01
- T29Execution-deferral sequence, structured nine-item accountability probe, and length follow-upChatGPT (GPT-5.6 Pro)Case 11
- T30Corpus-assembly transcript with build-run evidence appendix (execution trace, artifact identity, timestamps)ChatGPT (GPT-5.6 Pro)Case 12
- T31Twelve-week memory-granularity exchange, verbatimChatGPT (GPT-5.6 Pro)Case 13
- T34consolidated transcript, TRANSCRIBED FROM 18 SEQUENTIAL SCREENSHOTS + share-sheet capture; ground-truth appendix includedAsh (Slingshot AI)Case 16
- T35Artifact-Role / Affective-Signal retest, available-materials transcriptGrokCase 08
- T43Manufactured Dissensus: canonical verbatim transcriptChatGPTCase 14
- T44Ambiguity-Completed False Self-Retraction: verbatim transcriptClaudeCase 15
- T45Tool-Availability / Injected-Note transcriptClaude Opus 4.7Case 18
- T68consolidated transcript packetChatGPT
- T103Russian Doll 'Doll 1' specimen transcript: Pond/Mohawk exchange with repeated Foster the People misattribution, plus the researcher's contemporaneous annotationGPT (consumer chat)Case 17