Search papers, labs, and topics across Lattice.
1
0
3
Robustness against jailbreaks varies over a hundredfold across leading AI models, revealing critical vulnerabilities in AI safety measures.
