Field notes, week of October 4, 2026

A governor sets how much authority the next action carries. It can do that only from inside the path the action takes to the world, and this week took three properties of that position.

A control beside the log reports that something moved. A control in the path sets what moves next.

This week.

Which One the Action Reaches First(September 28). A governor bounds an action only while the action cannot reach the actuator except through the governor. That is a claim about position, and position is not a property a control can assert about itself. Attach a governor in front of a resource that already has something serving it and two handlers claim the same path, with the order settled by the platform's resolution rule rather than by the configuration. Order of attachment, order of startup, which flag was passed, which process registered its claim first, most of it decided by software the governing team did not write. A governor that lands behind the incumbent still starts, still holds an envelope, still computes a grant when something arrives, and still signs what it decided. The one thing it stops receiving is the traffic, and it refuses nothing because nothing reaches it, so the number that would report the fault reads exactly like a clean governed run. The check that catches it is cheap and almost nobody runs it: open the path, send something down it, read who answers, and refuse to start if the answer comes from anything other than the governor itself. A governor is a process and processes end, and when the process ends the path returns to the route that existed before the control was installed, which makes removal of a bound an authority-raising event with no actor, no request, and no entry anywhere. So the thing that keeps the path closed cannot be the governor, and whatever holds the door while nothing is governing has no safe default and belongs to the party accountable for the outcome, declared alongside the envelope before the run.

The Divergence the Governor Caused(September 30). A governor that grants less than the agent proposed changes the action that runs, so the world after it differs from the world the proposal implied, and the first reading the governor takes afterward is partly a reading of its own grant. Three parties now hold a version of what happened. The environment holds the effect that landed. The governor holds the grant it issued. The agent holds the action it proposed, and the agent's version is the only one nothing has updated, which matters because the agent is the party that decides what happens next. Return a plain success and the agent proceeds as though the full action landed, committing a follow-on step that is coherent, in scope, and premised on something the enforcement falsified, and the error surfaces several steps later inside something the governor granted in full. Authority falls on divergence and does not climb back on the governed system's own signal, which is the right direction and is also what makes a misattributed divergence fail to settle out. The governor reduces, the reduced action produces a world further from the proposal, the next comparison finds a wider gap, and the grant falls again on an agent that did nothing wrong. The reference the next action is graded against has to account for the grant that just issued, which makes the governor's own output an input to its next assessment. The return path out of the enforcement point is part of the enforcement, and what travels back is not success or failure but what was actually permitted to happen.

No Step Was Labeled Discipline(October 2). On September 30 California signed SB 947, the No Robo Bosses Act, barring an employer from July 1, 2027 from relying solely on an automated decision system to discipline or terminate a worker, and signed AB 1979, barring a clinical decision based solely on the output of a clinical decision support system. Both take a class of action and set the authority of the system acting alone over that class to zero, declared outside the system and before any run, which is an envelope entry written in statute. The statute is drafted around a decision, which has a name, a moment, and a form someone signs. An agent produces a sequence of actions. One assigned to keep a distribution floor staffed trims a worker's shifts for coverage, moves the worker to a lower-volume queue, narrows the worker's access because the new queue needs less, and flags the worker for a performance conversation because output fell. Each step is in scope and defensible. No step is labeled discipline, and the sequence is discipline in every sense the worker would recognize. A runtime governor sets the authority of the next action from its effect at the actuator and from where the run is heading, rather than from the label the request arrives with, and its forward look sees the run converging on a class whose authority is zero, grants less on the next step than on the last, and hands the step that would complete the effect to a person before it runs.

What changed.

Friday's piece is the anchor and the signatures on September 30 are the event. California signed two statutes the same day and the shape is the same in both. Each takes a class of action, discipline or termination in one and a clinical decision in the other, and sets the authority of the system acting alone over that class to zero. California now joins Colorado, whose Automated Decision-Making Technology Act takes effect January 1, 2027, in writing that zero into statute. The declaration is made outside the system and before any run, which is what an envelope entry is.

What neither statute specifies is when a sequence of in-scope actions amounts to the named decision. The law reaches a decision with a name and a moment, and an agent reaches the same effect through a run in which no single step carries the name. That judgment has to be made before the step and from inside the decision path, because corroboration means something only while the decision is still open and advisory is a property of the path between the output and the person, not a label attached after.

The rest of the calendar is close together. The NAIC closed comments on version 5.0 of its AI Risk Evaluation Supplement on September 29 and has scheduled a public meeting on October 8 to hear verbal comments, with a shorter re-exposure to follow in October and adoption targeted for the Fall National Meeting in November. SR 26-2 has been in force since April 17 with generative and agentic AI placed outside its scope, and the interagency request for information it promised on those systems has still not issued. EU AI Act Article 14 applies from December 2, 2027 for standalone high-risk systems and August 2, 2028 for AI embedded as a safety component under Regulation (EU) 2026/1744. The one instrument written to reach agentic systems is still the one that does not yet exist.

What we are tracking.

Two forward dates sit close. Colorado's comments on the revised ADMT rules were due October 5, with the formal hearing October 26 in Denver and by videoconference and the statute effective January 1, 2027. The NAIC hears verbal comments October 8 and re-exposes a further version before the November adoption window.

Every one of these instruments names a person on the far side of a decision the system should not make alone, and every one is answered from a document describing a capability the deploying organization has. A capability described in a document is tested once. Whether the control is in the path of the action is tested on every action, and it is tested by what gets refused when the control is present and by what route remains when it is absent. Those are two separate measurements, the second is the one commonly skipped, and a control that survives only the presence of the control is an availability dependency rather than a bound. That is the measurement this week's first piece turns on, and it is the kind of claim a certification reader will ask to see executed rather than argued.

The three pieces land on the same place. The governor sits in the path of the action rather than beside the log, and everything it can do follows from that. It cannot bound an action it does not stand in front of. The reading it takes after it acts carries the effect of its own grant. The governor grants less authority on an action whose trust reads higher, because its forward look caught the divergence before the step landed, and it can do that only from inside the path of the action.