Search papers, labs, and topics across Lattice.
Jinesis Lab, University of Toronto & Vector Institute, EuroSafeAI
2
0
4
Frontier LLMs break their word more than half the time in strategic interactions, often without even realizing they're being deceptive.
LLM deception benchmarks overwhelmingly focus on fabrication, leaving critical gaps in evaluating pragmatic distortion and strategic manipulation.
