Search papers, labs, and topics across Lattice.
1
2
Safety intentions in AI agents can degrade over time, leading to unsafe actions and operational failures, but a new architectural layer can intercept these issues effectively.
