# Methodology

## Release identity

- Report: AI Change Management Signal Report 2026
- Dataset version: 1.0.0
- Release date: 2026-08-04
- Corpus: 83 records (39 documented; 44 research)
- Observation period: 2026-02-16 through 2026-04-29
- Canonical report: https://michaelhbm.com/research/ai-change-management-signal-report-2026
- Frozen JSON distribution: https://michaelhbm.com/downloads/research/ai-change-management-signal-report-2026/v1.0.0/signal-corpus.json
- Canonical record-array SHA-256: b049fcb4b7b4ce980b112b2690060450cb1bea26a5c47bbd144f1dc1c3adb037

## Purpose

AI Change Desk reviewed dated AI platform, policy, privacy, security, governance, workflow, infrastructure, and distribution changes for their operating implications. The report is a practitioner editorial observatory designed to help leaders connect external change to adoption, authority, evidence, and rollback decisions.

## Unit of analysis

One record represents one dated operating signal. A signal records the change, Michael Hanna-Butros Meyering's operating analysis, one or more non-exclusive classification labels, and at least one public source relationship. Documented records cite the published AI Change Desk episode file in which the implication was developed. Research records link to a publicly accessible source being evaluated by the editorial desk. The public-source label describes access and citation, not whether a source is primary, independent, or authoritative.

## Inclusion rule

A record had to identify a dated AI platform, policy, privacy, security, governance, workflow, infrastructure, or distribution change with an operating implication. Records without a date, an operating analysis, a valid classification label, or a public source relationship were excluded from this release.

## Classification

Each record was classified against six non-exclusive operating lenses. A record may carry more than one label.

- **Governance (governance):** Can we swap vendors, export evidence, and explain our risk tiers under pressure?
- **Access (access):** Who can let an AI system act, and who can pause it immediately?
- **Deployment (deployment):** Are we choosing for speed, or for controllable continuity?
- **Security (security):** Who owns triage, patch approval, operator communication, and rollback in the same workflow?
- **Validation (validation):** What has to pass before a model or agent change is allowed to scale?
- **Ecosystem (ecosystem):** Are we buying software, or committing to an operating model?

## Counting

Theme counts are the number of distinct records carrying that label. Theme shares use all 83 records as the denominator. Because labels overlap, the 180 theme assignments do not sum to 83, and theme shares do not sum to 100 percent. Combined findings use a record-level union, so a record carrying multiple relevant labels is counted once.

## Lane definitions

- **Documented:** An operating implication already developed in a published AI Change Desk episode file.
- **Research:** A source-linked editorial lead still under evaluation; inclusion is not endorsement or a claim of independent verification.

## Limitations

This is a curated practitioner editorial corpus, not a representative survey of governments, employers, vendors, workers, products, or the AI market. Counts describe only this release and should not be generalized as prevalence estimates. Research-lane inclusion is not endorsement or independent verification. Sources can change or disappear after publication; the package preserves citations and analysis, not copies of third-party source materials.

## Frozen release versus living tracker

This v1.0.0 package is immutable and remains the citable evidence behind the 2026 report. The [AI Change Desk change tracker](https://michaelhbm.com/AiChangeDesk/changes) and its [machine feed](https://michaelhbm.com/AiChangeDesk/discovery/changes.json) are living editorial products and may add, revise, or reclassify signals. Future frozen corpora must use a new versioned directory and checksum manifest; v1.0.0 must not be overwritten.

## Reproducibility and integrity

Records are sorted deterministically by date descending, lane ascending, and title ascending. JSON uses two-space indentation and a final newline. CSV uses UTF-8, RFC 4180-style quoted fields, a pipe-delimited theme field, and a compact JSON source field. Verify files against **manifest.json** or **SHA256SUMS** using SHA-256.