Guidelines  ·  2026-09-22

Proof-of-Control v1.0 working draft — open standard for verifiable AI agent control

GuidelinesMedium impactGlobal
On 17 September 2026, the Advanced AI Society released the v1.0 working draft of 'Proof-of-Control' for public comment through October 30, 2026, now hosted as a lab at the Linux Foundation's LF Decentralized Trust. Co-designed with 80+ enterprise security leaders (review board includes Bruce Schneier), the open standard defines tamper-evident, openly-verifiable evidence that an AI agent stayed within the controls it was given — verifiable by any neutral third party without trusting the operator. It addresses the 'Verifiability Gap' as autonomous agents act faster than human oversight, and responds to bipartisan Congressional proposals demanding continuous agent verification.
Agent execution verifiability is an emerging control requirement with no incumbent standard; Proof-of-Control is one of the first open standards attempting to make AI agent control-governance openly auditable rather than operator self-attested. Hosting at LF Decentralized Trust gives it an open-governance home, and it is explicitly framed around statutory mandates now moving through Congress. A public comment period is open, so practitioner and vendor input can still shape v1.0 before it hardens.
Organisations building or deploying agentic AI should review the v1.0 draft and submit comments before 30 October 2026, and evaluate whether tamper-evident, third-party-verifiable control evidence belongs in their agent procurement and audit requirements.
Advanced AI Society — Proof-of-ControlTNW — Advanced AI Society joins Linux Foundation, launches Proof-of-Control
See this in the live feed Explore related AI security and governance findings — updated every morning.
Open the feed →