Search papers, labs, and topics across Lattice.
5
0
5
22
A single manipulated search result can dramatically amplify the effectiveness of attacks on LLM-based search agents, revealing critical vulnerabilities in their evidence-gathering processes.
A safety trilemma reveals that LLMs cannot balance utility, safety, and access when evidence is copyable, exposing a critical vulnerability in current safeguards.
FARI achieves faster and more robust watermark verification in diffusion models, outperforming traditional methods with a single inversion step.
Achieving over 89% accuracy in attributing generated images to their source models, DNA redefines the landscape of digital forensics in image generation.
Semantic guardrails can't stop AI agents from going rogue, but formal logic constraints might.
