Heron opens AI security research fellowship for cybersecurity professionals

The autumn 2026 program will match 10 to 12 research teams with AI security advisors to build prototypes, benchmarks, papers and technical reports.

Heron AI Security is recruiting experienced cybersecurity professionals for its autumn 2026 AI Security Research Fellowship.

Heron AI Security has opened applications for its autumn 2026 AI Security Research Fellowship, giving experienced cybersecurity professionals until July 8 to join research teams working on frontier AI security projects.

The fellowship will run from September to November 2026 and is expected to support 10 to 12 research teams. Heron is seeking cybersecurity professionals with at least five years of experience who can commit between eight and 30 hours a week.

The program is remote-first, with optional coworking hubs in London, Tel Aviv and San Francisco. Teams are due to be announced in early August, with projects scheduled to launch on September 1.

Heron says applicants do not need an AI background. The fellowship is aimed at people with experience in areas such as cryptography, AI and machine learning implementation, agent security, red-team operations, adversarial testing, secure systems, infrastructure design, proof-of-concept development and attack simulations.

The fellowship is unpaid, but Heron says all teams will receive more than $1,000 in compute, Claude Code Team accounts for the duration of the program and conference attendance support of up to $5,000 per team.

Research teams will work with AI security advisors

Heron describes the fellowship as a set of “collaborative projects at the intersection of Cybersecurity and frontier AI,” where cybersecurity professionals work with AI security field leaders on publishable and practical research.

Each team will include an expert advisor, two to four cybersecurity professionals, a project manager and a technical advisor. Heron says teams are expected to produce publishable results, open-source prototypes or technical reports by the end of the program.

The autumn 2026 projects cover AI infrastructure and hardware security, technical AI governance, adversarial and model security, AI control and containment, and cybersecurity evaluations and demos.

Project titles listed by Heron include Reproducible-Purpleteam, Computation-Aware, Training-Leakage, Dfir-Agents, Tiered-Disclosure, Containment-Redteam, Cross-Monitor, Backdoor-Inheritance, Insider-Honeypots, Encryptor-Risk, Enclave-Api, Training-Custody, Ot-Decoy and Cot-Leakage.

Advisors include researchers from AI, cyber and policy organizations

Named advisors and project leads include Brad Edwards, Architect and Researcher at Palo Alto Networks; Yossi Gandelsman, Scientist at Reve.art and Assistant Professor at TTIC; Gabriel Kulp, Fellow at the RAND Center on AI, Security, and Technology; and Zainab Majid, Co-Founder of Asymmetric Security.

The list also includes Sean McGregor, Co-Founder and Lead Research Engineer at AVERI and Fellow at Harvard University; Pingshan Zhang, GovAI Fellow and former Microsoft employee; Itay Yona, AI Researcher at Google DeepMind; and Nitzan Shulman, Head of Cyber at Heron.

Heron says the fellowship is designed to connect cybersecurity domain expertise with frontier AI context. The organization says: “Cybersecurity professionals bring domain expertise; field leaders bring frontier-AI context. Together, they build proofs-of-concept, benchmarks, or defensive strategies.”

The program schedule includes biweekly programming at either 15:00 to 16:00 UTC on Tuesdays or 18:00 to 19:00 UTC on Saturdays, plus weekly research meetings with experts, project managers and technical advisors.

Application window closes this week

The application process includes an application form, a work trial and a short interview. Heron says candidates will be matched to projects based on team fit.

Mid-project presentations and milestone submissions are scheduled for October 2026, with final submissions and a showcase event planned for November.

Heron also says teams will receive a guaranteed presentation at the AI Security Forum in Tel Aviv. Applications for the autumn 2026 fellowship close on July 8.