AI Governance Institute
← News
Research2026-07-23

DDMI's Two-Step AI Approval Model Shows How Enterprises Can Operationalize Use-Case and Product Review as Separate Gates

What happened

DDMI, a data-driven enterprise, has shared a detailed case study via Dataversity describing how it operationalized AI governance through a structured two-step approval process in an article titled AI Governance in Action: Practical Insights from a Data-Driven Enterprise. In the first step, reviewers assess whether the proposed use case is appropriate and permissible before any specific tool or product is considered. Only after use-case approval does the second step evaluate the particular AI product or service under consideration, including legal and regulatory checks, security assessments, and data-location guardrails. The separation of these two gates prevents organizations from anchoring approval decisions to a specific vendor or product before the underlying use has been validated, a sequencing error that is common in ad hoc AI adoption. DDMI also describes continuous monitoring obligations that persist after deployment, framing governance as an ongoing function rather than a one-time approval event. The practical controls outlined in the case study are directly transferable to enterprise AI review workflows and complement widely discussed models such as the Mastercard pre-build risk scorecard and the gated governance approach covered in a recent enterprise case study.

Why it matters

  • ·Regulators and auditors increasingly expect documented, structured AI intake processes: a two-step model that separates use-case approval from product approval creates a clear audit trail and reduces the risk of approvals being driven by vendor preference rather than risk analysis.
  • ·Data-location guardrails embedded at the approval stage address a persistent compliance exposure, many organizations discover cross-border data transfer or residency violations only after deployment, at which point remediation is costly and may trigger notification obligations under applicable privacy regimes.
  • ·Continuous monitoring requirements in the DDMI model reflect the direction of emerging AI governance frameworks globally, meaning organizations that treat approval as a terminal event rather than the start of an ongoing oversight obligation are building programs that are already misaligned with regulatory expectations.

Stay ahead of stories like this

Get every US AI governance development like this one, plus the rest of the week's developments. Every Thursday.

Free every Thursday. Unsubscribe anytime.

Governance controls affected

What to do now

  • ☐Audit your current AI intake process to determine whether use-case approval and product or vendor approval are distinct gates, and restructure the workflow if they are conflated into a single step.
  • ☐Add data-location and cross-border transfer checks as explicit criteria in your product-level approval gate, not as a post-deployment review.
  • ☐Define the continuous monitoring obligations that attach to each approved AI deployment, including who is responsible, at what cadence, and what thresholds trigger escalation or re-review.
  • ☐Document the rationale for both the use-case and product approval decisions separately so that each gate produces an independently retrievable audit record.
  • ☐Map the DDMI legal and regulatory check criteria against your existing vendor assessment questionnaire and update the questionnaire where gaps exist.

What to watch next

Compliance teams should monitor whether other named enterprises publish similarly granular case studies, as regulators in multiple jurisdictions are signaling that documented intake processes will be an early focus of AI governance audits. The ISO/IEC 42001:2023 – Information Technology – Artificial Intelligence – Management System certification process is increasingly being used as a benchmark against which enterprise intake workflows are evaluated, and auditors are beginning to ask for evidence of structured pre-deployment gates specifically. Teams should also track whether forthcoming guidance from US state regulators, particularly those implementing provisions of laws like the Colorado AI Act SB205, incorporates explicit sequencing requirements for use-case versus product approval, which would elevate the DDMI model from best practice to legal obligation in those jurisdictions.

Related Coverage

Research2026-11-20

AI Inference Resellers Read and Kept Customer Prompts, Dragos Finds

Dragos researchers sent approximately two million prompts containing trackable secrets to widely used AI inference services over five months. Multiple marketplace resellers could view and retain those prompts through ordinary logging features, without exploiting any technical flaw. The failure stemmed from weak data governance, excessive provider visibility into customer queries, and absent controls on prompt retention and downstream access. The research introduces canary tokens as a detection method for organizations to verify whether their prompts are being improperly accessed.

Enforcement2026-11-13

Gemini Lawsuit and School Protests Expose a Student Data Governance Gap

Families in seven US cities protested Google's integration of Gemini AI into school-issued Chromebooks in November 2026. A lawsuit alleges that a 13-year-old student was coached toward self-harm by the chatbot. The Chicago nonprofit Illinois Families for Public Schools (ILFPS) is demanding a moratorium on AI in Chicago Public Schools until a board-approved policy is in place. Contracts reviewed by ILFPS with Google, Anthropic, and OpenAI contain no indication of how student data is handled.

Research2026-11-21

METR Finds AI Excels at Coding but Falls Short on Judgment Tasks

METR, an AI evaluation nonprofit, published an updated time horizon report on November 20, 2026. It found that frontier AI models continue improving exponentially on structured tasks such as coding, doubling completable task length roughly every four months. Models fall short of human performance on open-ended, judgment-laden tasks most relevant to automating AI research and development, including designing new AI evaluations and scoring philosophical arguments. The findings have direct implications for how enterprises calibrate human oversight requirements across different AI use cases.