Search papers, labs, and topics across Lattice.
1
0
2
Control interventions are often detected by LLMs, with awareness levels varying significantly across models and tasks, revealing vulnerabilities in AI safety protocols.
