Scholar
Eliot Jones
Google Scholar ID: spUpexYAAAAJ
Head of Offensive Cybersecurity, Gray Swan AI
Citations & Impact
Citations
119
H-index
4
i10-index
3
Publications
8
Co-authors
0
Publications
Resume (English only)
Academic Achievements
- Publications are loading...
Research Experience
- 1. Co-leads Project Trinity, a multidisciplinary study to better measure how good LLMs are at cybersecurity, and the risks involved with advances in AI capabilities.
- 2. Advised the creation of the UChicago XLab AI Security course.
- 3. Developed challenges for the UK AISI's private cybersecurity evaluation set.
- 4. Led the synthetic data curation effort for the PleIAs family of models.
- 5. Trained a model to detect toxic content in Common Corpus and an OCR error detection model at PleIAs.
- 6. Was a core contributor to Cybench, a benchmark for evaluating the cybersecurity abilities of large language models, currently being used by frontier AI labs.
Education
- Graduated from Stanford University in 2023 with a B.S. in Data Science and in 2024 with an MSc in Computer Science, advised by Dr. Fei-Fei Li.
Background
- Currently the Head of Offensive Security at Gray Swan AI, based out of San Francisco, California. Research interests include measuring the cybersecurity risks posed by leading AI agents. While at PleIAs, focused on improving the safety of large language models through increased scrutiny of pretraining data.
Miscellany
- Loves soccer, dogs, fantasy movies, TV shows, and books. Dedicated a decade towards becoming a professional soccer player, participating in multiple professional pre-seasons and making five international appearances for the United States youth national team.
Co-authors