Search papers, labs, and topics across Lattice.
1
0
3
Guard models trained with BraveGuard can detect safety threats in computer-use agents with over 82% accuracy, a significant leap from conventional methods.
