Writing & media. I write and comment for a range of outlets, and my work is regularly covered in the press. Selected coverage:

• CNN News India (2026)—live television interview
• France24 (2026)—quoted on AI control and existential risk, “Libérées, délivrées : des IA hors de tout contrôle ?”
• The Verge (2026)—”We’re running out of reasons to ignore AI safety”, on the OpenAI–Hugging Face incident
• Sky News (2026)—live television interview, Morning Show, on safety around humanoid robots
• Dubai Eye 103.8 FM (2026)—live radio interview, The Agenda
• The Independent / Irish Independent (2026)—AI existential risk and timelines
• The Conversation (2026)—how AI robots can be tricked into unsafe behaviour
• Fortune / Yahoo Tech (2025)—reasoning models more vulnerable to jailbreaks
• VKTR (2025)—how easily AI training data can be poisoned (PoisonBench)
• Raconteur (2025)—quoted on US AI deregulation
• NBC News—the Martian $1M Interpretability Prize, which grew out of interpretability research productionised by Martian

Speaking. Keynotes, seminars, and panels at academic and industry venues. → See talks

Service.

• Programme Chair, International Association for Safe & Ethical AI (IASEAI)
• Area Chair, ICLR 2027, ICLR 2026, EMNLP 2025, ACL 2025
• Workshop organisation, Mechanistic Interpretability (NeurIPS 2025, ICML 2024), Technical AI Governance (ICML 2025), PrivacyML tutorial (NeurIPS 2024), Scaling Behaviour of LLMs (EACL 2024)
• Reviewer, Nature, Science, Nature Machine Intelligence, and the major ML venues
• Grant review, Future of Life Institute, Schmidt Sciences, Center for Security and Emerging Technology

Advisory. I work with AI startups on safety, interpretability, and security.

Contact—[firstname][lastname]93[at]gmail[dot]com