Defensive poisoning followed by unlearning erases most unknown backdoors in LLM tool-calling agents, while trigger recognition can outlive malicious execution.
Verbalizable representations form a global workspace in language models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Backdoor Decontamination Dynamics in LLM Agents
Defensive poisoning followed by unlearning erases most unknown backdoors in LLM tool-calling agents, while trigger recognition can outlive malicious execution.