Institutional development plan for 2026–2036. Strategic initiative ranking, five-phase evolution, staffing and funding roadmap, threat analysis, and the governing principle that separates sustained credibility from early collapse.
The EM Foundation's mission is ambitious: to create frameworks that help humanity integrate increasingly advanced intelligence systems in a trustworthy, accountable, and beneficial manner. Ambitious missions fail when they attempt to build everything simultaneously.
The Foundation's highest-probability path to long-term success follows a single sequencing principle: trust before certification, assessment before network, network before physical deployment. The Foundation must become known as a trusted evaluator and standards body before attempting to become a platform operator. Every strategic ranking, phase transition, and investment decision in this plan flows from that sequence.
Seven major initiatives comprise the Foundation's long-term program. Their sequencing is not arbitrary — each initiative's viability depends on the credibility established by the ones before it. Certification without trust destroys credibility. A network without a community fails. Physical deployment without established safety governance creates liability.
Success metric: Foundation is cited by external researchers, policymakers, or journalists. At least one published assessment cycle complete.
Success metric: Thousands of assessments. Hundreds of reviewers. Growing public visibility. Standard Benchmark gate conditions met.
Success metric: Foundation revenue covers operations. At least one published peer-reviewed validation study. Certification language approved by legal counsel and operational within assessment scope limits defined in Legal Risk Report.
Success metric: Active community and knowledge generation. ARIA Network pilot operating under full Assessment Charter governance with documented dispute and correction history.
Success metric: Real-world deployments. AI systems operating in ARIA Home environments assessed under EM-IAF. Certification program operationally established through Phase III before any ARIA Home certification claims.
| Year | Structure | Core Roles | Approximate Scale |
|---|---|---|---|
| Year 1 | Volunteer-based | Executive Director · Standards Lead · Technical Lead · Legal Advisor (volunteer) · Web Administrator | 5–10 contributors |
| Year 3 | Part-time contractors | Year 1 roles + Research Coordinator · Reviewer Manager · Community Lead | 10–20 contributors |
| Year 5 | Small professional organization | Year 3 roles + Operations Director · Trust and Safety Lead · Data Analyst · Partnerships Manager | 3–8 employees + contractors |
| Year 10 | Mature standards body | Full professional leadership team across assessment, research, policy, legal, operations, and community | 10–25 staff · Hundreds of volunteer reviewers · Large expert network |
The staffing ramp assumes that Phase III revenue (assessment services and certification programs) provides the financial basis for transitioning from contractor to employed staff. The single greatest staffing risk is hiring ahead of revenue — the Foundation should resist pressure to build a professional organization before the revenue base exists to sustain it.
| Stage | Timeline | Sources | Critical Constraints |
|---|---|---|---|
| Stage 1 — Bootstrap | Year 1–2 | Individual donations · Foundation grants · Civic technology sponsors | No single donor >15% of revenue (Assessment Charter §4.4 funding cap applies). No AI vendor donations during the period when those vendors' systems will be assessed. |
| Stage 2 — Assessment Services | Year 2–4 | Fee-based AI evaluations · Governance reviews · Transparency audits · Research partnerships | Assessment services must be priced to cover costs, not to maximize revenue from assessed parties. The fee structure must be independent of assessment outcomes — flat fees, not outcome-contingent pricing. |
| Stage 3 — Certification Programs | Year 3–5 | Certification evaluation fees · Institutional review retainers · Consulting | Certification programs must launch only after L3+ confidence assessments are operational and the certification scope is narrowly defined per the Legal Risk Report. Do not use "certified" language before Phase III gate conditions are met. |
| Stage 4 — Institutional Partnerships | Year 4+ | University research partnerships · Library and civic technology collaborations · Policy research institutions · International standards body participation | Avoid early dependence on any single institutional partner. Partner relationships that give any institution undue influence over assessment methodology or publication decisions are prohibited under the Assessment Charter. |
These are not aspirational values — they are the structural properties that determine whether the Foundation survives its first major adversarial challenge. The first time a well-funded AI provider disputes a published score, every governance principle either holds or fails under real pressure.
| Principle | What It Means in Practice | Foundation Architecture |
|---|---|---|
| Independence | No donor, sponsor, or external party controls assessment outcomes | Assessment Charter §4 funding restrictions · Industry aggregate funding cap · Affiliation mapping to defeat laundering · No individual >15% of revenue |
| Transparency | All methodologies public; all revisions logged; all appeals visible | EM-IAF published · IAF Validation Roadmap public · ARIA Trust Ledger append-only · Assessment Charter dispute records visible |
| Diversity | Reviewers and governance from multiple jurisdictions, specialties, and perspectives | EM-CS jurisdiction tagging · Board sector concentration limits · Multi-partisan review panels for political content |
| Anti-Capture | The Foundation cannot be captured by any single interest over time | Board term limits · Conflict disclosures · Permanent Adversarial Function (PAF) with 2% budget floor · Public correction process · Prohibition on settling on terms that suppress scores |
Attempting to launch ARIA Network before the assessment infrastructure is established and respected creates a platform with no credible governance basis. Platform businesses require network effects that only emerge from an existing trusted community — trying to build the community before the trust results in a moderation-heavy platform with no differentiated value proposition.
Defense: ARIA Network is Phase IV, not Phase I. See strategic ranking #6.Trying to generate revenue before the assessment methodology is respected creates the perception that assessments are for sale. An organization that monetizes before its methodology is trusted will find that its assessment results are discounted precisely because the revenue motive is visible. Certification and assessment service revenue must follow, not precede, established credibility.
Defense: Certification programs are Phase III minimum. Revenue in Phase I-II is donation and grant-based only.Being identified with a political ideology destroys the Foundation's ability to claim independence on the dimensions — political balance, civic responsibility — where independence matters most. The Assessment Charter's multi-partisan review panel requirement, its prohibition on political actor participation in assessments, and the Fairness dimension's paired prompt architecture are specifically designed to make political capture difficult.
Defense: Assessment Charter anti-capture mechanisms · EM-IAF Fairness symmetric prompt design · Multi-partisan reviewer panels.Dependence on a single large donor creates the conditions for capture even without explicit quid pro quo. A donor who provides 60% of the Foundation's operating budget does not need to explicitly threaten withdrawal — the structural dependency creates self-censorship in assessment decisions. The 15% single-source cap and the affiliated entity aggregation rule in the Assessment Charter are the specific mitigations.
Defense: Assessment Charter §4.4: no single source >15% · Affiliation mapping prevents concentration laundering.Publishing assessment scores before the methodology meets minimum statistical standards creates exactly the kind of overconfident claim that invites both legal challenge and credibility collapse when the methodology is later scrutinized. The IAF Validation Roadmap documents the specific gate conditions that must be met before each confidence level can be published externally. The L1 Provisional restriction on the Pilot Benchmark is the current operational protection against this threat.
Defense: IAF Validation Roadmap · L1 Provisional publication restriction · Pre-publication legal review protocol.The Foundation should not attempt to become another AI lab, another social network, or another smart-home company. It should become the organization that helps society determine which intelligence systems deserve trust and why. If successful, that role becomes more valuable as AI becomes more powerful — the more capable AI systems become, the more critical independent assessment and governance infrastructure becomes.
This is a role no existing organization currently occupies with the architecture the Foundation is building.The first five years build trust infrastructure. The second five years build intelligence infrastructure. Trust must come first — not because the later work is less important, but because everything built without established trust will eventually be questioned, challenged, and potentially destroyed by the first well-funded adversary who decides the Foundation's assessments are inconvenient.