Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
3
0
4
Even when individual audits clear EEG models, they can still leak sensitive spectral attributes, revealing a critical gap in current evaluation methods.
Brain-prompt injections can manipulate BCI-LLM agent actions undetected, exposing critical vulnerabilities in current safety protocols.
VLA models can't have it all: there's a fundamental limit to how much capability and robustness you can squeeze out of them, and this paper proves it.