ADDED: Governance & Compliance terms from GRR WG. - #2
Conversation
Add 11 entries to taxonomy-data.js (28 -> 39) under the approved Governance & Compliance category: Accountability, Agent deactivation, Agent oversight, Audit trail, Autonomy level, Escalation pathway, Explainability, Guardrails, Human on the loop (HOTL), Kill switch, and Least privilege (agent). Apply reciprocity edits (relatedTerms/contrastsWith only) to six existing entries so cross-references remain bidirectional: HITL, Autonomous, Deterministic policies, Delegated authorization, Blast radius, Hooks and checkpoints. Terms cross-referenced against NIST AI RMF, EU AI Act, OWASP Agentic Top 10, ISO/IEC 42001, Berkeley CLTC, and the Singapore Model AI Governance Framework for Agentic AI (2026).
|
Some information rather than a review, since the terms themselves are for the Domain Editors. This branched from 720e0ec on 7 June and main has moved 44 commits since, so it no longer merges. The reason is worth knowing before any rebase. On 13 June, commit 56c50d0 deliberately emptied the array, removing all 28 terms so that each could be re added through a consistent path. This PR predates that. The description reads 28 to 39 terms, and measured against its own base the diff is 129 added lines, but measured against main today it is 431, because merging it would also restore the 28 terms that were intentionally cleared. So the 11 Governance and Compliance terms here still look wanted, and they all sit in the approved One thing worth catching either way. The alias @kjamithash happy to help with the split if that is useful. |
Add 11 entries to taxonomy-data.js (28 -> 39) under the approved Governance & Compliance category: Accountability, Agent deactivation, Agent oversight, Audit trail, Autonomy level, Escalation pathway, Explainability, Guardrails, Human on the loop (HOTL), Kill switch, and Least privilege (agent).
Apply reciprocity edits (relatedTerms/contrastsWith only) to six existing entries so cross-references remain bidirectional: HITL, Autonomous, Deterministic policies, Delegated authorization, Blast radius, Hooks and checkpoints.
Terms cross-referenced against NIST AI RMF, EU AI Act, OWASP Agentic Top 10, ISO/IEC 42001, Berkeley CLTC, and the Singapore Model AI Governance Framework for Agentic AI (2026).