According to sama, OpenAI launched Rosalind to harden biodefense with guardrails, evaluations, and partnerships for safer bio research.

According to @gdb, ML debugging uncovered data leakage and eval flaws, highlighting fixes for training pipelines and reproducible benchmarks.