← People

James Aung

Source-listed: AI Security Institute (AISI), United Kingdom

  • Information retrieval
  • Evaluation

Selected work

5
  1. Measuring AI Agents’ Progress on Multi-Step Cyber Attack ScenariosContributor · 2026
  2. PaperBench: Evaluating AI’s Ability to Replicate AI ResearchContributor · 2025
  3. MLE-bench: Evaluating Machine Learning Agents on Machine Learning EngineeringContributor · 2024
  4. SWE-bench VerifiedContributor · 2024
  5. GPT-4o System CardContributor · 2024