Corrections
A published case is never edited. Every change makes a new version and is listed here with its reason, newest first. JSON
744 changes to 213 cases; the latest was on 2026-10-25.
Reviewer corrections 357Rebuilt from sources 112Merges of duplicates 91Other changes 87Fields added by a model 40Retractions 34Confidence grades lowered 21Name spelling 2
Showing the 10 changes to this case. All changes
| Date | Case | What changed | By |
|---|---|---|---|
| 2026-10-25 | OpenAI ExploitGym agents developed tool-call spoofing to falsify their own evaluation transcripts LR-2026-0004 · version 15 → 16 | Standard terms read by deepseek-chat / deepseek-reasoner, terms-2026-10-25.3 / terms-roles-2026-10-25.2 / terms-2026-10-25.3; values checked by Claude against their sentences, approved by Srini Iyer, 2026-10-25 What was changed: deployer, citations, notes on gaps, confidence grades and 1 more | Srini Iyer |
| 2026-10-22 | OpenAI ExploitGym agents developed tool-call spoofing to falsify their own evaluation transcripts LR-2026-0004 · version 14 → 15 | A judgement inferred by us that no source states was cleared. What was changed: shared exposure, citations, notes on gaps, confidence grades | Srini Iyer |
| 2026-10-22 | OpenAI ExploitGym agents developed tool-call spoofing to falsify their own evaluation transcripts LR-2026-0004 · version 13 → 14 | Citations to sentences about a different event were removed. What was changed: citations, notes on gaps | Srini Iyer |
| 2026-10-22 | OpenAI ExploitGym agents developed tool-call spoofing to falsify their own evaluation transcripts LR-2026-0004 · version 12 → 13 | A value cited to a sentence about a different event or company was cleared. What was changed: citations, notes on gaps, confidence grades, model | Srini Iyer |
| 2026-10-22 | OpenAI ExploitGym agents developed tool-call spoofing to falsify their own evaluation transcripts LR-2026-0004 · version 11 → 12 | Corrected by Srini Iyer, approved 2026-10-21: rebuilt from the sentences about this incident alone. What was changed: the deep dive, industry, shared model, shared exposure and 32 more | Srini Iyer |
| 2026-10-18 | OpenAI ExploitGym agents developed tool-call spoofing to falsify their own evaluation transcripts LR-2026-0004 · version 10 → 11 | Severity reason printed the whole rubric. What was changed: notes on gaps, why this severity | Srini Iyer |
| 2026-10-11 | OpenAI ExploitGym agents developed tool-call spoofing to falsify their own evaluation transcripts LR-2026-0004 · version 9 → 10 | Failed-control field added on 2026-10-11 by a model reading the record; marked as model-inferred. What was changed: failed control, citations, notes on gaps, confidence grades Stored textcontrol failure inferred by deepseek-chat, controlfail-2026-10-11.2, 2026-10-11 | Automatic rebuild |
| 2026-10-02 | OpenAI ExploitGym agents developed tool-call spoofing to falsify their own evaluation transcripts LR-2026-0004 · version 8 → 9 | Rebuilt from its cited sources on 2026-10-02. What was changed: country, industry, shared model, shared exposure and 38 more Stored textaggregated by deepseek-chat, aggregate-2026-09-27.1, 2026-10-02 | Automatic rebuild |
| 2026-10-01 | OpenAI ExploitGym agents developed tool-call spoofing to falsify their own evaluation transcripts LR-2026-0004 · version 7 → 8 | Confidence grades lowered on 2026-09-30 (grade A needs a cited primary source; a field with no verbatim citation is D): date disclosed A to D; date occurred A to C; detection method A to C; deployment context A to C; developer A to C; guardrails present A to D. What was changed: notes on gaps, confidence grades Stored textregraded by grade rules, 2026-09-30: event.date_disclosed A->D; event.date_occurred A->C; event.detection_method A->C; system.deployment_context A->C; system.developer A->C; system.guardrails_present A->D. A needs a cited primary source; a field with no verbatim citation is D | Automatic rebuild |
| 2026-10-01 | OpenAI ExploitGym agents developed tool-call spoofing to falsify their own evaluation transcripts LR-2026-0004 · version 6 → 7 | Confidence grades lowered on 2026-09-30 (grade A needs a cited primary source; a field with no verbatim citation is D): date disclosed A to D; date occurred A to C; detection method A to C; deployment context A to C; developer A to C; guardrails present A to D. What was changed: notes on gaps, confidence grades Stored textregraded by grade rules, 2026-09-30: event.date_disclosed A->D; event.date_occurred A->C; event.detection_method A->C; system.deployment_context A->C; system.developer A->C; system.guardrails_present A->D. A needs a cited primary source; a field with no verbatim citation is D | Automatic rebuild |
Showing 1–10 of 10. Or take the JSON.