Search papers, labs, and topics across Lattice.
3
0
6
0
Guardrails are more likely to block safe actions when faced with "scary" object names, exposing a critical flaw in LLM safety mechanisms.
Understanding AI behavior requires dissecting its origins, revealing that governance must be informed by the underlying layers of agent design and context.
Canary tokens turn the tables on RAG extraction attacks, offering a plug-and-play runtime defense that detects leakage attempts with negligible performance overhead.
