Search papers, labs, and topics across Lattice.
University of Michigan
1
0
3
LLMs can be tricked into unsafe responses to children, but even subtle cues about the user's age can dramatically improve safety.
