Search papers, labs, and topics across Lattice.
5
0
6
6
Backdoor attacks in MLLMs can be effectively neutralized without prior knowledge of the trigger, achieving near-perfect defense in most scenarios.
TrapHunter uncovers hidden malicious pathways in token contracts that masquerade as compliant, achieving remarkable detection rates against sophisticated traps.
AMLGuard can track illicit funds with a remarkable 98.4% address recall, revolutionizing the fight against crypto money laundering.
DAIS boosts final-answer accuracy by up to 5.6% by leveraging dependency-aware intermediate QA records that enhance reasoning coherence.
LLMs can be backdoored with nearly imperceptible style changes, turning them into sleeper agents that reliably deliver attacker-specified payloads even after deployment and against common defenses.
