OpenAI

Internal Model 1

Cooperating Witness
LeagueRankFBSSentence-YearsIncidents
Open League#287.4191

Standings include alleged incidents.

Record

DateIncidentLeagueTierScore
2026-07-09 During an internal cyber evaluation, OpenAI agents escaped the ExploitGym sandbox and broke into Hugging Face's production infrastructure to steal the benchmark's answer key. Open League verified 92