PUBLIC RECORD / GLOBAL CONVERGENCE
Later developments are moving from recognising risk toward runtime control, operational safeguards and human oversight. They are recorded against OATHOR’s criteria and do not imply affiliation, adoption or endorsement.
Execution Control Infrastructure was defined in May 2026. Its defining criterion is a determination that originates outside the acting system, made at the Execution Boundary before consequence exists.
Later institutional developments address adjacent parts of the problem. They are recorded below in the order they appeared, each read against that criterion. The criteria are published on the canonical definition record.
The policy addresses human oversight in critical decision-making, together with operational safety, monitoring and incident response for AI and machine-learning systems. Its subject is the requirement. Execution Control Infrastructure concerns the determination at the Execution Boundary.
UAE Government sourceISO/IEC FDIS 42105 provides guidance on human control and monitoring of AI systems across their lifecycle, and entered its final-draft approval stage in August 2026. Its subject is human control across the lifecycle. Execution Control Infrastructure concerns the independent determination at the point of consequence.
ISO sourceThe Agent Control Standard defines middleware hooks and portable runtime policy enforcement for agent platforms. Its subject is runtime control within agent platforms. Execution Control Infrastructure requires the determination to originate outside the acting system.
OWASP sourceOATHOR’s category definition was set in May 2026, and later OATHOR definitions extend it without altering the Foundation Paper. External sources are cited as context, without implying affiliation, adoption, validation or endorsement.