Percy Liang
- Audits Under Resource, Data, and Access Constraints: Scaling Laws For Less Discriminatory Alternatives
- Blackbox Model Provenance via Palimpsestic Membership Inference
- BountyBench: Dollar Impact of AI Agent Attackers and Defenders on Real-World Cybersecurity Systems
- Establishing Best Practices in Building Rigorous Agentic Benchmarks
- MLE-Dojo: Interactive Environments for Empowering LLM Agents in Machine Learning Engineering
- Machine Unlearning Doesn't Do What You Think: Lessons for Generative AI Policy and Research
- On the Entropy Calibration of Language Models