Updated October 2026
Methodology & sources
How the workforce AI security requirements are verified and mapped: verification methods, pilot measures, framework editions, sources and scope limitations.
Disclaimer
Mappings indicate relevant contributions, not complete compliance, certification, assurance, a conformity assessment or legal advice. Read each explanation and its scope conditions. Legal applicability depends on jurisdiction, role, use and effective dates. Reference summaries are selective; consult current official sources and full licensed standards before relying on them.
Verification methods
- LT
- Live test
- POV
- Proof of value
- W
- Walkthrough
- D
- Documentation
- PM
- Pilot measurement
Pilot measures
| ID | Measure | How | Why |
|---|---|---|---|
| PM-01 | Detection rate by data class | Share of sensitive content in a reviewed sample of pilot prompts correctly flagged, per data class, on default configuration. | Shows whether detection understands your data, not just generic identifiers. |
| PM-02 | False-alert proportion | Share of reviewed flagged pilot events judged benign: benign flagged events divided by all reviewed flagged events. Record the review window and sample size. Report not measurable if no flagged events were reviewed. | The best predictor of whether employees and the SOC will tolerate the control. |
| PM-03 | Inline decision time | Time added to a prompt submission when a decision is made before it is sent. | Decides whether real-time enforcement is practical. |
| PM-04 | Language coverage | Detection rate for pilot prompts in each language your workforce uses. | A control that only reads English leaves much of the workforce unprotected. |
| PM-05 | Surface coverage | Policy outcomes observed across browser, desktop, command line, coding agent and MCP during the pilot. | Exposes gaps between what is claimed and what is enforced. |
| PM-06 | Behavior change | Share of coached interventions where the employee edits or cancels, and the trend in repeat exposures. | Shows whether the control changes behavior or just records it. |
Framework editions
- SOC 2 / AICPA 2017 Trust Services Criteria with revised points of focus 2022
- ISO/IEC 27001:2022 including Annex A, with Amd 1:2024 noted
- ISO/IEC 42001:2023
Sources
- S1 AICPA 2017 Trust Services Criteria with Revised Points of Focus 2022 — AICPA publication page for the Trust Services Criteria and revised points of focus.
- S2 AICPA 2022 redline of the Trust Services Criteria — AICPA criteria text, including criterion identifiers, meanings and introductory guidance. Redline deletions are not current requirements.
- S3 AICPA SOC Suite of Services — Overview of AICPA SOC assurance services.
- S1 ISO: ISO/IEC 27001:2022 publication and overview — Publisher overview of ISO/IEC 27001:2022 and the ISMS certification scope.
- S2 ISO: ISO/IEC 27002:2022 overview — Publisher overview of ISO/IEC 27002 guidance; it is not a certification standard.
- S3 ISO/IEC 27001:2022 public preview, GSO-hosted — Public preview of ISO/IEC 27001 clause headings and scope. Consult the full standard for requirements.
- S4 ISO/IEC 27002:2022 public preview, GSO-hosted — Public preview of ISO/IEC 27002 control headings. Consult the full standard for implementation guidance.
- S5 ISO/IEC 27001 Auditing Practices Group: Statement of Applicability note — Educational guidance on risk-selected controls and the Statement of Applicability; not normative ISO text.
- S6 ISO: ISO/IEC 27001:2022/Amd 1:2024 — Publisher page for the 2024 amendment.
- S7 IAF and ISO: climate amendment explanation — Joint IAF/ISO explanation of climate-related considerations in organizational context.
- S9 ISO/IEC 27001:2022 public sample hosted by iTeh — ISO-authored public sample covering risk assessment, risk treatment and the Statement of Applicability.
- S8 IEC: ISO/IEC 27001:2022 — IEC co-publisher page for ISO/IEC 27001:2022.
Renumbered IDs
IDs were renumbered. Old IDs still work as links (for example /r/DIS-001) and redirect to the current requirement.