Search papers, labs, and topics across Lattice.
2
0
4
Only one out of 37 open-weight AI models meets the necessary evaluation standards to ensure safety and robustness, revealing a critical oversight in current practices.
Evaluation design choices can drastically alter the perceived risks of AI systems in biological research, revealing a hidden complexity that demands attention.
