Search papers, labs, and topics across Lattice.
4
0
7
0
Benign experiences in self-evolving LLMs can be weaponized, revealing a hidden attack surface that undermines safety guarantees.
IG-GAN slashes aerodynamic data generation errors by up to 97% by harnessing the power of intrinsic geometry.
LLM-based multi-agent systems are surprisingly vulnerable: a new RL-based attacker can evolve sophisticated, long-horizon attacks by exploiting trust in external tools.
Audio deepfake detectors trained on diffusion-reconstructed "hard" examples generalize far better to unseen attacks, slashing error rates compared to standard training.
