Mentoring

I love mentoring research projects. If you are interested in working with me on threat modelling, AI control, honeypots for AI agents or related topics, the best route is through one of the programmes below — or just reach out.

Opportunities to work with me

• Applications open: MATS Winter 2027 — I will co-mentor with Sambhav Maheshwari on threat modelling and control for national-security deployments of AI agents.

Mentored projects

• Vidya Gopalakrishnan, Sudhiksha Kandavel Rajan, Vishnu Nair & Yadnyesh Chakane (Heron Security Fellowship, ongoing): Evaluating honeypots as a control technique against malicious AI agents
• Rocio Perales Valdes (IAPS Fellowship, ongoing): Threat modelling how advanced robotics could contribute to catastrophic risk
• Manon Kempermann & Max Ramackers: Fine-Tuning: From Benign to Malign — How Different Fine-Tuning Methods Degrade Safety Guardrails in LLMs During Benign Fine-Tuning