Publications

2025

1. Paper

   Evaluating the Goal-Directedness of Large Language Models

   Tom Everitt, Cristina Garbacea, Alexis Bellot, and 4 more authors

   2025

   arXiv Bib

   @article{everitt2025evaluatinggoaldirectednesslargelanguage,
     title = {Evaluating the Goal-Directedness of Large Language Models},
     author = {Everitt, Tom and Garbacea, Cristina and Bellot, Alexis and Richens, Jonathan and Papadatos, Henry and Campos, Siméon and Shah, Rohin},
     year = {2025},
     eprint = {2504.11844},
     archiveprefix = {arXiv},
     primaryclass = {cs.AI},
     url = {https://arxiv.org/abs/2504.11844},
   }

2. Paper

   Mapping AI Benchmark Data to Quantitative Risk Estimates Through Expert Elicitation

   Malcolm Murray^*, Henry Papadatos^*, Otter Quarks, and 2 more authors

   2025

   arXiv Bib

   @article{murray2025mappingaibenchmark,
     title = {Mapping AI Benchmark Data to Quantitative Risk Estimates Through Expert Elicitation},
     author = {Murray, Malcolm and Papadatos, Henry and Quarks, Otter and Gimenez, Pierre-François and Campos, Simeon},
     year = {2025},
     eprint = {2503.04299},
     archiveprefix = {arXiv},
     primaryclass = {cs.AI},
     url = {https://arxiv.org/abs/2503.04299},
   }

3. Paper

   A Frontier AI Risk Management Framework: Bridging the Gap Between Current AI Practices and Established Risk Management

   Simeon Campos^*, Henry Papadatos^*, Fabien Roger, and 3 more authors

   Conference on frontier AI safety frameworks (2024), 2025

   arXiv Bib

   @article{campos2025afrontierairisk,
     title = {A Frontier AI Risk Management Framework: Bridging the Gap Between Current AI Practices and Established Risk Management},
     author = {Campos, Simeon and Papadatos, Henry and Roger, Fabien and Touzet, Chloé and Quarks, Otter and Murray, Malcolm},
     year = {2025},
     eprint = {2502.06656},
     archiveprefix = {arXiv},
     primaryclass = {cs.AI},
     url = {https://arxiv.org/abs/2502.06656},
     journal = {Conference on frontier AI safety frameworks (2024)},
   }

2024

1. Report

   Rating Frontier AI Developers’ Risk Management Maturity

   Henry Papadatos, Simeon Campos, and Malcolm Murray

   Conference on frontier AI safety frameworks, 2024

   PDF Website
2. Paper

   Linear Probe Penalties Reduce LLM Sycophancy

   Henry Papadatos, and Rachel Freedman

   SoLaR workshop, NeurIPS, 2024

   arXiv Bib

   @article{papadatos2024linearprobepenaltiesreduce,
     title = {Linear Probe Penalties Reduce LLM Sycophancy},
     author = {Papadatos, Henry and Freedman, Rachel},
     year = {2024},
     eprint = {2412.00967},
     archiveprefix = {arXiv},
     primaryclass = {cs.AI},
     journal = {SoLaR workshop, NeurIPS},
   }

3. Blog post

   Your LLM Judge may be biased

   Henry Papadatos, and Rachel Freedman

   2024

   Blog