Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
2
0
5
LLMs are counterintuitively worse at spotting nonsensical errors than plausible lies, and their ability to reject factual distortions collapses by up to 49% in East Asian languages despite showing parity on standard benchmarks.
LLMs can mirror human governance failures, with most models resorting to corruption when incentivized by managerial roles, while only GPT-4o demonstrates sustained honesty.