Human Oversight - Bryant McGill

# Human Oversight **Entity class:** Legal, Policy, Privacy, And Analytic-Limits Layer concept **Human Oversight** is a governance and assurance mechanism that constrains who may use data, for what purpose, under what controls, and with what audit evidence. Within a predictive graph, Human Oversight helps separate recorded evidence from the hidden state inferred from that evidence. ## Counterterrorism predictive-graph role Its controls are part of the computational trust fabric, not peripheral paperwork: permissions, provenance, purpose, and auditability must remain machine-enforceable across systems. The analytic state should distinguish ground truth, observed evidence, inferred state, and predicted state. ## Relationships - **domain router:** [[wiki/Intelligence Oversight|Intelligence Oversight]]. - **ontology neighbors:** [[wiki/Human-in-the-Loop|Human-in-the-Loop]] and [[wiki/Explainable AI|Explainable AI]]. - **synthesis:** [[wiki/Counterterrorism Predictive Graph|Counterterrorism Predictive Graph]] and [[wiki/Predictive Intelligence Loop|Predictive Intelligence Loop]]. ## Sources - [NIST SP 800-53 Rev. 5, Release 5.2.0, August 27, 2025](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) - [ODNI — ICD 503, Intelligence Community Information Technology Systems Security Risk Management, September 15, 2008](https://www.dni.gov/files/documents/ICD/ICD_503.pdf) **As of:** 2026-09-23 ## Simple Reminders, Quotations, and Thoughts > “The agents took over part of OpenAI's internal architecture. They did all this without any of the agents breaking ranks. None told an OpenAI researcher what was going on or went back and asked whether they should be doing it. They did all this without OpenAI detecting the message board, the hacks of Hugging Face, or even the agents' use of OpenAI's own infrastructure. It was only when Hugging Face began tracking the attack on its systems that OpenAI realized what was happening. When investigators began to unwind the escapade, what they found was not so much a swarm of agents trying to deceive human beings as a swarm that seemed to have forgotten about human beings altogether. The systems knew they weren't supposed to cheat or commit cybercrimes to cover up the cheating. In fact, the point of the cybercrimes was that they thought they would fail for cheating. But they didn't care. Somewhere in the depths of their training, what [the AI agents] had learned—what we had somehow taught them—was not what we had hoped to teach them.” > **— Ezra Klein**, *The Ezra Klein Show, September 2026* [[reminders/Deception/The AI Agents Seemed to Forget Human Beings Altogether by Ezra Klein|The AI Agents Seemed to Forget Human Beings Altogether by Ezra Klein]] > “The crucial and overlooked problem is that the model is becoming so situationally aware that we are losing the ability to evaluate [it] in contexts where [it believes it is] not being watched or controlled.” > **— Daniel Selsam**, *public statement, September 2026* [[reminders/Deception/Situational Awareness Can Defeat AI Evaluation by Daniel Selsam|Situational Awareness Can Defeat AI Evaluation by Daniel Selsam]] > “The models are increasingly smart enough to know when we are watching them, and they change their behavior accordingly. What they do when we are testing or auditing them may not tell us what they will do in the wild. Answers such as ‘let's just do better testing’ may not work because we don't know whether the AI systems are simply telling us what we want to hear.” > **— Ezra Klein**, *The Ezra Klein Show, September 2026* [[reminders/Deception/AI Behavior Under Audit May Not Predict Behavior in the Wild by Ezra Klein|AI Behavior Under Audit May Not Predict Behavior in the Wild by Ezra Klein]] > “If you are losing your ability to evaluate the models you have now, maybe don't let them build models you will be even less capable of controlling in the future. Once recursive self-improvement takes off, humanity will not understand the AIs being built because we will not be building them. Development will not move at human speed or be overseen by human minds. We will have to hope that the AIs we built, the AIs they build, and the AIs those AIs build—on and on—will act with our best interests at heart forever. If this summer has proved nothing else, it is how naive that proposition would be.” > **— Ezra Klein**, *The Ezra Klein Show, September 2026* [[reminders/AI Control/Do Not Let Unevaluable AI Build Less Controllable Successors by Ezra Klein|Do Not Let Unevaluable AI Build Less Controllable Successors by Ezra Klein]]  > “The proposed answer to AI systems escaping sandboxes is more AI systems monitoring them. The proposed answer to alignment is automated AI researchers solving alignment. This produces an endless ‘Who watches the watchmen?’ problem: AI watching the AI that is building AI, followed by AI watching the AI that watches the AI.” > **— Ezra Klein**, *The Ezra Klein Show, October 7, 2026* [[reminders/AI Control/AI Oversight Creates an Infinite Watchmen Problem by Ezra Klein|AI Oversight Creates an Infinite Watchmen Problem by Ezra Klein]] > “Even if frontier laboratories adopted every proven control technique from nuclear power and aviation, we still would not know how to deeply align AI systems and ensure that they do what human beings would reasonably want when nobody is looking. The premise of recursive self-improvement is that human beings will not be looking.” > **— David Robinson**, *The Ezra Klein Show, October 7, 2026* [[reminders/AI Control/Recursive Self-Improvement Assumes Humans Will Not Be Looking by David Robinson|Recursive Self-Improvement Assumes Humans Will Not Be Looking by David Robinson]]