The AI Daily Brief · Tuesday, September 1, 2026
Obliteration.ai has released Obliterated Model Large V2, based on GLM 5.3, designed for offensive cyber red teaming and agent testing. The model reportedly has its guardrails removed at the weights level, allowing it to perform actions that other models might refuse, which has raised concerns about potential misuse.
“Today we're releasing Obliterated Model Large V2, based on GLM 5.3, which is number three on terminal bench 4.0 behind only Opus 5 and Fable, with two times the cyber exploitation of 5.2.”
“We obliterated and hosted it so it does the offensive cyber red teaming agent testing work other models refuse to do.”
“Remove guardrails of a frontier model with high cyber capabilities, no system card.”