Fully funded AI Security Bootcamp opens applications for London program

The seven-day program is aimed at experienced cybersecurity professionals and will cover frontier AI threats, LLM security, infrastructure risk and AI governance.

AI Security Bootcamp 2026 is accepting applications for a fully funded London program focused on frontier AI security, threat modeling, LLM security and infrastructure risk.

AI Security Bootcamp has opened applications for its 2026 London cohort, a fully funded seven-day in-person program for experienced cybersecurity professionals looking to work on frontier AI security.

The program will run in London from August 30 to September 5, 2026, with applications closing on July 15. AI Security Bootcamp says applications are being reviewed on a rolling basis and early applications are encouraged.

The cohort will be limited to 16 to 20 participants. Tuition, meals during program hours, materials and accommodation in London will be covered for accepted applicants, with need-based travel support also available.

The bootcamp is designed for professionals with at least five years of hands-on security experience across areas such as offensive security, incident response, threat intelligence, infrastructure security and application security. Prior AI or machine learning experience is not required.

Participants will complete pre-work before the program to cover baseline machine learning fundamentals, followed by a week of demos, lectures, guest speakers and hands-on red and blue team exercises.

Curriculum covers LLM and infrastructure security

AI Security Bootcamp’s 2026 program will focus on the threat landscape around frontier AI systems, from current enterprise deployments to risks linked to more capable models.

The first day covers threat modeling, including current misuse risks, application security, infrastructure security, model theft and tampering, integrity attacks, misalignment and governance guarantees.

Later sessions will cover adversarial examples, attacks on image models, trojans, backdoors, fine-tuning attacks, model weight extraction, watermarking, training data protection and inference-time data handling.

The LLM security section includes jailbreaks, prompt injection, RAG injection, guardrails, tokenization vulnerabilities and Model Context Protocol security.

Infrastructure content includes NVIDIA Container Toolkit exploits, GPU isolation, confidential computing, sandbox design, data center infrastructure, personnel security, hardware supply chains and governance frameworks.

Selection process includes technical assessment

The application process has three stages. The first stage is a short CV and application review, with AI Security Bootcamp saying it will look at signals including GitHub, past projects and CV evidence.

The second stage is a technical assessment covering threat modeling, a small PyTorch exercise and a brief Python exercise. The third stage is a 30-minute interview covering the applicant’s background, motivation and short technical questions.

Selection will prioritize candidates interested in frontier AI risk, high-consequence failure modes or work involving sophisticated threat actors. Experience with deep learning frameworks such as PyTorch is listed as a plus, but not required.

AI Security Bootcamp says accepted participants must attend the full program from Sunday, August 30 to Saturday, September 5. Participants are advised to arrive on Saturday, August 29 and depart on Sunday, September 6.

AI Security Bootcamp says past participants have been affiliated with organizations including OpenAI, Google, Meta, Apple, Microsoft, AWS, Intel, Jane Street, Darktrace and the Center for AI Safety.

The program page also names academic and research institutions including Stanford University, the University of Oxford, the University of Cambridge, MIT, UC Berkeley, Tel Aviv University, CERN, the National University of Singapore and the Technion: Israel Institute of Technology.