Search papers, labs, and topics across Lattice.
1
0
3
LLM-based recommender systems can trigger users' personal trauma, phobias, or self-harm history, but a new framework cuts these safety violations by 96.5% while maintaining recommendation quality.
