Theory of Change
In a Gist
The AI race is a global coordination failure that voluntary commitments cannot solve. We target the highest-leverage node: the dozen individuals who will actually shape Trump's AI policy. Our 350-page Strategic Memo maps each influencer's philosophy, interests, and persuadable predictions. Our October 2025 Tour validated the approach: 85+ contacts, 23 AI lab officials, direct pathways to 2 of 10 primary targets—all on $72,000. The anticipated Trump-Xi summit in April 2026 creates a time-bound window. We are the only organization focused entirely on this leverage point. If P(treaty attempt) shifts even modestly, expected value is astronomical.
The Problem
Target Audience: This Theory of Change is written for highly knowledgeable long-term AI safety and governance activists and experts. If you are not, you may want to refer to our Case for Concerned Citizens.
The race toward artificial superintelligence is a global coordination failure. Even AI lab leaders who acknowledge catastrophic risk—Amodei assigns 10-25% probability to extinction, Musk 20%—continue racing because unilateral restraint merely hands advantage to competitors. As our Strategic Memo v2.6 documents (pp. 49-50, "Collapsing Timelines"), timelines to transformative AI are compressing while governance mechanisms remain frozen in multilateral gridlock.
The trajectory leads to one of two catastrophes: loss of control (unaligned ASI emerging from an uncoordinated race) or authoritarian capture (AI concentrated in unaccountable hands). Neither UN processes nor voluntary commitments can move fast enough to matter.
The Intervention
We target the highest-leverage node in the AI governance landscape: the dozen individuals who will actually shape Trump's AI policy—and through Trump, any realistic near-term US-China coordination.
Our 354-page Strategic Memo profiles each key influencer across three dimensions: interests (how a treaty serves their position), philosophy (where their worldview creates openings), and AI predictions (which probability estimates can be shifted). See the detailed framework in "The Five-Part Analysis Framework" (pp. 170-175) and individual profiles for Amodei (pp. 226-247), Vance (pp. 180-205), Pope Leo XIV (pp. 274-295), and others.
The Causal Chain
Step 1: Build Introducer Networks → Step 2: Secure Influencer Access → Step 3: Shift Probability Estimates → Step 4: Create Coalition → Step 5: Unified Pitch to Trump
Our October 2025 US Tour validated Steps 1-2: we generated 85+ contacts (vs. 15-20 projected), including 23 AI lab officials and direct introducer pathways to 2 of 10 primary targets. This demonstrated that credibility transfers—not cold outreach—unlock influencer access.
Step 3 exploits a key insight from our analysis (see "Why Influencers Can Be Substantially Swayed," Memo pp. 38-47): most target figures are driven by philosophical commitments, not financial interests. Vance genuinely fears technofeudalism. Bannon truly opposes oligarchic concentration. Amodei authentically worries about the catastrophic risks he's helping create. When people act from conviction rather than calculation, they can change their minds—if shown their deepest commitments are better served by treaty advocacy.
Step 4 leverages the three-camp structure we've identified (Memo pp. 41-42): Conservative Humanists (Vance, Bannon, Pope), Techno-Humanists (Amodei, Hassabis, Sacks), and Trans/Post-Humanists (Altman, Musk, Thiel). The humanist camps form a natural majority. Coalition-building among them creates social proof that accelerates individual persuasion.
Addressing Key Objections
"Won't this lead to authoritarian capture?" We address this extensively in "Treaty Enforcement that Prevents both ASI and Authoritarianism" (Memo pp. 130-138). The paradox of mutual distrust between Trump and Xi is a feature: for a treaty to be credible to both, it must constrain both. China has strong incentives to push for structures preventing any single nation from dominating—including itself (see "China's Paradoxical Interest in Democratic Governance," pp. 125-126).
"Can treaties actually work?" The Baruch Plan passed the UN Atomic Energy Commission 10-0 in 1946. It failed at the Security Council, but every failure mode—slow diplomacy, inadequate verification, narrow political coalition—is now addressable. See "Could Better Treaty-Making Infrastructure Have Saved the Baruch Plan?" (pp. 116-119).
Evidence of Traction
Our 2025 achievements on $72,000 total funding (~$180/high-value meeting) demonstrate capital efficiency unmatched in this space. We've secured testimonials from experts across UN, NSA, WEF, Yale, and Princeton. Our 2026 Roadmap projects 150+ introducer engagements and 5-8 substantive influencer meetings before the anticipated Trump-Xi summit in April 2026.
The Expected Value Calculation
If P(treaty attempt given intervention) shifts even modestly, and P(treaty preventing catastrophe | attempt) is non-trivial, the expected value is astronomical given the stakes. We are the only organization focused entirely on this leverage point.