The report defines AI integrity threats (model sabotage and subversion) and recommends four US government policy actions to defend frontier AI systems against backdoors and secret loyalties.
Subliminal Learning: Language Models Transmit Behavioral Traits via Hidden Signals in Data
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CY 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
AI Integrity: Defending Against Backdoors and Secret Loyalties
The report defines AI integrity threats (model sabotage and subversion) and recommends four US government policy actions to defend frontier AI systems against backdoors and secret loyalties.