Operating agenda
- Define auditable deployment gates for frontier models.
- Build a shared evidence standard for evals, incident reports, and capability monitoring.
- Recruit young talent through replication tasks and falsifiable threat models.
Problem DetailProblems
Frontier AI audit stack
Turn AI safety into measurable engineering: evaluations, monitoring, red-team data, governance, and independent audit trails.
DomainAI Safety
Score98
Next proofAnti-gaming eval suite