Making online worlds harder to weaponise

Led Project Catalyst, a large-scale, rapid consortium of 8 implementing partners running 12 pilots across 4 countries, tackling violent extremism, online misogyny and TFGBV. Supervised novel AI classifier development, advised the Foundation's AI safety programme on adversarial misuse of AI models.

Facilitated ministerial-level dialogue on reintegrating families of former ISIS-affiliated individuals returning from Al Hol camp, trained 50+ senior officials and wrote strategic communications plans.

Red-teamed generative AI models as an external expert, defining typologies of model risk exploited by violent extremist and terrorist actors to inform model-safety strategy.

Wrote Discord's violent extremism policy and internal enforcement guidelines, working across trust & safety, engineering and policy teams.

Authored policy and safeguarding guidance on protecting children from organised violence, grooming and exploitation in online gaming communities.

Wrote Prevent, Detect, React, a framework for countering violent extremism on gaming surfaces, as a member of GIFCT working groups.

Supervised multi-regional research and grant programmes for eight years, from the impact of COVID-19 on extremist narratives to recruitment, reintegration and AI spaces in Southeast Asia.

Led strategic communications for Indonesia's flagship violence prevention programme: seven national-scale studies that shifted government P/CVE policy, and 44 trainings for 880+ participants whose campaigns reached more than 4 million people.

As Research & Insights Director, led 100+ applied research products across a 30-project Asia-Pacific portfolio on violent extremism, conflict, hate speech and mis/disinformation, for clients including Meta, TikTok, YouTube, UNDP, and USAID.