What happened
On 11 October 2026 a new individual Internet-Draft 'Rule Provenance for Autonomous AI Agent Audit Records' (draft-piehl-agent-rule-provenance-00, P. Piehl, Experimental) was submitted to the IETF datatracker. It specifies an experimental profile of the Agent Audit Trail (AAT) format that records the rules an agent operates under — each rule version, who set and approved it, and which rules were in force when a decision was made — separates decisions from later approvals/overrides, binds them to the exact request, and requires consequential-action records to go to an independent recorder anchored in a public log.
Why it matters
It extends the emerging Agent Audit Trail logging format (draft-sharif-agent-audit-trail, already tracked) with forensic rule provenance so operators, courts and regulators can verify whether an AI agent was allowed to act. This is a concrete step toward auditable agentic-AI accountability standards.
Action needed
Teams building agent audit logging should watch this profile as a candidate extension to AAT for rule/policy provenance in compliance and incident-response contexts.