Specific theory of change:

Phase 1: Create policy supply (Months 1-6)

• Launch AI Policythons at 20 universities across 10 countries, focusing on institutions near policy centers (DC, Brussels, Geneva, Tokyo)
• Each policython produces 5-10 briefs on specific LAWS regulation mechanisms: verification protocols, dual-use export controls, liability frameworks, red lines for autonomy levels
• Target output: 150+ policy briefs covering the full regulatory stack

Phase 2: Policy distribution (Months 3-12)

• Partner with existing organizations (Future of Life Institute, Campaign to Stop Killer Robots, ICRC) to route briefs to decision-makers
• Briefs get distributed to: UN CCW delegates, EU AI Act working groups, national defense committees, and NGOs building the treaty coalition
• Success metric: 30+ briefs cited in official proceedings or policy papers

Phase 3: Movement building (Months 6-18)

• Policython participants become ongoing advocates, similar to how Model UN creates future diplomats
• Build a network of 500+ policy-literate AI safety advocates embedded in government, think tanks, and advocacy orgs
• Create a talent pipeline into organizations actively working on LAWS regulation

Why this works:

1. Speed: Weekend events can be launched globally in months, not years
2. Proven model: I've done this before with Policython.org - our research was adopted by Opportunity Insights and influenced $20B+ in COVID policy
3. Fills a gap: Treaty campaigns need specific, implementable policy proposals. Right now there's ~10 organizations doing advocacy but insufficient policy R&D capacity
4. Scalable: Student organizers can host independently after initial support, like how I helped launch policythons in Toronto and Philippines
5. Network effects: Creates a global community of policy-engaged AI safety people who continue working on this after the events

Concrete next steps: