The AISM Scoring Update, released October 2, 2026, fixes a confusion that has been quietly corrupting AI security decisions: a good product is not a mature organization, and a strong claim is not strong evidence. The update clarifies the existing AISM Sovereignty Score and introduces two new review instruments, the Technology Contribution Profile v1.0 and Evidence Assurance v1.0, so that organizational maturity, a technology's scoped contribution, and the assurance of each claim are scored separately and never folded into one another.
That separation is the whole point. A promising product, paper, or benchmark can address a control gap without establishing your organization's maturity. Security leaders need to see what the technology does, where it works, and which evidence supports the claim. Before this update, those three questions could be answered with a single number that proved none of them.
The problem: one score that proves nothing
Consider the decision a security leader actually faces. A vendor demonstrates a tool that closes a control gap. A research paper shows a benchmark result. A benchmark score looks impressive. The question is not whether the tool is good. The question is whether this candidate addresses the organization's actual control gap, and what remains uncovered.
A single maturity score cannot answer that. Product capability, test evidence, and organizational maturity are different things, and when they are conflated, each one borrows credibility from the others. The product looks mature because the organization is. The organization looks mature because the product is strong. The claim looks proven because the evidence was never examined on its own. The update exists to stop that borrowing.
What changed: three instruments, kept apart
The release keeps the existing AISM Sovereignty Score and clarifies it: organizational maturity across five pillars and six dimensions, kept separate from technology contribution and claim-level evidence assurance. It then adds two instruments that do the jobs the score was never meant to do.
| Instrument | What it measures |
|---|---|
| AISM Sovereignty Score | Organizational maturity across five pillars and six dimensions |
| Technology Contribution Profile v1.0 | What an artifact addresses, implements, enforces, validates, evidences, or challenges, for named controls within a stated boundary |
| Evidence Assurance v1.0 | The evidence supporting a specific attributed claim, with strength, origin, independence, and limitations recorded |
The before-and-after is direct. Before this update, organizational scoring had no dedicated technology-review entry point; now a separate profile records control effects, coverage, evidence, boundaries, and residual risk. Before, product capability, test evidence, and maturity could be conflated; now the review distinguishes the three. Before, new research needed translation into concrete adoption decisions; now nine source-attributed examples frame validation work, complementary controls, and ownership.
The Technology Contribution Profile is the working instrument. It asks a reviewer to record, for a specific artifact at a specific version and deployment boundary, what the artifact addresses, implements, enforces, validates, evidences, or challenges. The reusable Technology Card captures version, scope, controls, mechanisms, evidence, failure behavior, retained authority, and adoption conditions in one place. The validation guidance covers clean and attacked pairs, side-effect and bypass tests, policy-integrity checks, reconstruction, and notification ownership. The incident evidence workflow defines the next review steps when something goes wrong.
What the result means, and what it does not
The release gives each instrument a plain meaning. The Sovereignty Score is organizational maturity across five pillars and six dimensions. The technology contribution is the capability demonstrated for named controls within the stated boundary. Evidence Assurance is the evidence supporting a specific attributed claim, with independence recorded separately. Where the review has not established a score or assurance rating, the result is "not assessed," and missing evidence stays visible instead of being averaged away.
Two rules in the release deserve emphasis, because they are the ones most likely to be broken in practice. First, a technology contributes evidence; an organization earns its maturity level. Do not label a product "AISM Level 4" or treat released code as independent validation. Second, the new assurance rubric stays separate from existing CLI grades and Challenge Lab levels. Existing CLI E0-E5 numeric grades cannot be relabeled as assurance v1.0 ratings.
The release also states its own boundaries plainly. It documents how to review control mechanisms and evidence. It does not install an enforcement system, independently validate the example technologies, or certify an organization. Runtime integrations, new experiments, and Technology Contribution Profile CLI and dashboard support remain backlog proposals. An incident notification or subpoena is not proof of a breach or technical effectiveness. Tests must state their actual platform, tool paths, enforcement planes, bypasses, and exclusions.
What to do with it
No installation is required. Start with the AISM assessment guide, which covers the three instruments, the Technology Card workflow, and a worked example. The workflow has three steps.
First, identify the control gap and the artifact's exact version and deployment boundary, using the assessment workflow in the Technology Contribution Profile. Second, copy the Technology Card and record each control effect, supporting artifact, coverage, uncertainty, and failure mode. Third, review the bounded ContractWarden design example, then apply the adoption rule with a named decision owner and acceptance conditions.
One boundary governs the whole exercise: completing a card does not authorize execution, deployment, external notification, or a conformance claim. The documented tests are proposals, not experiments performed by this update. The card is a review instrument. The decision it supports still belongs to a named human.
The validation behind it, stated with its limits
The hosted checks passed for the methods pull request's final source revision 7cf6477, including the Python 3.11 to 3.14 CLI and AISM matrix, NEXUS tests, Markdown lint, repository and manifest consistency, security gates, and CodeQL analysis. The follow-up guidance, method-status, and template update in the second pull request also passed all 28 runnable hosted checks at reviewed revision 446e29a.
Those checks validate the repository changes. They do not validate the external examples. The release says so directly, and this announcement does not claim more. The nine review candidates turn supplied research and accountability examples into scoped acceptance work, without adopting unsupported scores.
What stays the same matters too. Framework and CLI versions, organizational scoring, existing CLI evidence weights, schemas, native tool scores, and frozen Challenge 001 remain unchanged. AI SAFE2 retains 161 core controls; MCP retains 19 profile controls, and UAS retains 27 profile requirements. The documentation names the model the AI Sovereign Maturity Model while preserving historical records and distinct instrument names.
The next steps are also stated as work, not as capabilities. Complete artifact-specific cards, assign experiment and incident-review owners, and collect the evidence needed for adoption decisions. Future machine schemas, CLI commands, adapters, and dashboard views require separate implementation and validation.
Start with one control gap and complete a Technology Card for the candidate. That is the release's own instruction, and it is the right one. A single card, done honestly, tells a security leader more than a maturity score ever did, because it keeps the three questions apart: what the organization can do, what the technology contributes, and what the evidence actually proves.
Sources
- AISM Scoring Update, October 2, 2026: https://github.com/CyberStrategyInstitute/ai-safe2-framework/releases/tag/2026-10-02_AISM_Scoring_Update
- AISM assessment guide: https://github.com/CyberStrategyInstitute/ai-safe2-framework/blob/446e29a51c3dff75d3b4def9ce9a49c6cb13f077/AISM/ASSESSMENT-GUIDE.md
- Technology Contribution Profile v1.0: https://github.com/CyberStrategyInstitute/ai-safe2-framework/blob/446e29a51c3dff75d3b4def9ce9a49c6cb13f077/docs/TECHNOLOGY-CONTRIBUTION-PROFILE.md
- Evidence Assurance v1.0: https://github.com/CyberStrategyInstitute/ai-safe2-framework/blob/446e29a51c3dff75d3b4def9ce9a49c6cb13f077/docs/EVIDENCE-ASSURANCE.md
- Technology Card template: https://github.com/CyberStrategyInstitute/ai-safe2-framework/blob/446e29a51c3dff75d3b4def9ce9a49c6cb13f077/docs/templates/TECHNOLOGY-CARD.md
- Nine review examples: https://github.com/CyberStrategyInstitute/ai-safe2-framework/blob/446e29a51c3dff75d3b4def9ce9a49c6cb13f077/research/technology-contribution-examples.md
- AISM model README: https://github.com/CyberStrategyInstitute/ai-safe2-framework/blob/446e29a51c3dff75d3b4def9ce9a49c6cb13f077/AISM/README.md
- Release notes audit, November 17, 2026: ~/workspace/seo-review/release-notes-audit-2026-11-17.md