AI Security Bootcamp San Francisco, January 2027

The AI Security Bootcamp is a seven day intensive course for security professionals who want to work on securing frontier AI systems. The January 2027 cohort runs in person in San Francisco from January 10 to 16, 2027.

The course is free. Tuition, meals during programme hours, materials and accommodation in San Francisco are covered for accepted participants, and travel support is available based on need.

Each day has its own topic: threat modelling; adversarial attacks, watermarking and data security; LLM security; infrastructure security; weight security, verification and formal methods; data centre security; and AI control and hardware governance. Teaching mixes demos, lectures, guest speakers and hands on red and blue team exercises.

Applicants need at least five years of hands on security experience. No AI or ML background is needed. Selection favours people interested in risks from frontier AI, failures with serious consequences or work involving advanced threat actors. Participants must attend all seven days and do the reading set before the course.

Applications close on November 7, 2026 (anywhere on earth) and are reviewed on a rolling basis, so the organizers encourage early applications.

Quick answers

Do I need a machine learning background?

No. The pre-work covers the ML basics. Experience with PyTorch is a plus but not required.

How many people take part?

The San Francisco cohort has 20 places.