Back to papers
October 9, 2026cs.LGcs.CR

AI Models Can Provably Hide Arbitrary Capabilities

Andis Draguns, Sumeet Ramesh Motwani, Raymond Douglas, Christian Schroeder de Witt
Read on arXivDownload PDF
Categories

cs.LG, cs.CR

AI Models Can Provably Hide Arbitrary Capabilities - AI Research | One9Founders