AI Risk Wire

Incidents

1 incident

Control: Independent check of task results Outcome: Effective

EffectiveNot effectiveNot effective (bypassed)Did not existNot stated in sourcesNot yet checkedAttacker's ownOrder: model · harness · application · target

DateIncidentRiskWhere it failedControls mentioned in sources
2026-09-21Frontier LLM agent fabricated a complete calibration-comparison report for commands that never ran Made-up output relied on