The Liat Show: A Story Unfolding Across Timelines.
Entered through Portal
Current Time
AI governance has become much better at defining responsibility, oversight, and intervention.
What it still does not show clearly is whether the human reviewer formed an independent judgment or began from the system’s output.
This is the standalone Authority Drift Governance Checklist, designed to make that distinction visible. The full article explaining the governance framework and the reasoning behind these questions is available in Oversight Measures Presence. Not Legitimacy.
This checklist is for people responsible for AI governance who review whether AI is being used properly inside an organization. That includes governance professionals, risk committees, AI oversight teams, and internal auditors. This checklist does not replace existing AI governance frameworks or the requirements of NIST AI RMF, ISO/IEC 42001, the EU AI Act, or ISO/IEC 38507; those frameworks remain necessary. It extends them by adding the questions those frameworks do not currently ask about where judgment forms in practice.
Use this checklist when reviewing any AI system involved in consequential decisions.
On Judgment Formation
- Does the reviewer encounter the system’s output before or after forming an independent view of the problem?
- Is there a defined point in the review workflow at which the reviewer is expected to assess the situation independently, before consulting the system?
- If the reviewer’s independent view differs from the system’s output, is that difference documented?
- Is the reviewer able to reconstruct the reasoning behind a decision without referring to the system’s explanation?
On Override Behavior
- Are override rates tracked and reviewed over time?
- Has the override rate changed since the system was deployed, and in which direction?
- When an override occurs, is it documented with the reviewer’s reasoning?
- Does agreement with the system require less justification than disagreement?
- Are reviewers evaluated or rewarded in ways that create incentives to align with the system rather than to exercise independent judgment?
On Default Reference Behavior
- Has the system’s output become the starting point for team discussions in the relevant decision domain?
- Do team members begin by reading the system’s recommendation, or by forming their own assessment?
- Is the system treated as the answer to be confirmed or as one input among several?
- When the system is unavailable, are team members able to perform the relevant judgment independently?
On Legitimacy Signals
- Do people express a need for a specific reason to disagree with the system, beyond simply forming a different view?
- Is the system referred to as an authority rather than a tool in internal communications?
- Do new team members learn the domain primarily through the system’s outputs rather than through independent training?
- Is the system’s judgment treated as a baseline that human judgment must exceed to be taken seriously?
On Governance Design
- Does the current review process include a step in which independent judgment is formed before the system’s output is consulted?
- Are reviewers given access to the original evidence and data, independently of the system’s interpretation of it?
- Is the workflow designed to encourage verification of the system’s output, or confirmation of it?
- Would a reviewer reach the same conclusion if the system were not available?
On Agent Autonomy Specifically
- For agentic systems, does governance track not only whether the system stayed within its permissions, but also whether it has become the default source of judgment about what the next step should be?
- Is there a review mechanism for detecting when an agentic system’s recommendations are adopted without independent evaluation?
- Has the organization assessed whether the practical authority of any agentic system has shifted beyond its formally authorized scope, not through permission violation but through repeated deference?
What This Checklist Does and What Comes Next
The checklist is free to use and circulate. It helps governance teams identify which areas require closer investigation. It does not assess a specific organization or determine how its findings relate to existing controls. It also does not specify what evidence must be collected or how implementation should proceed.
The Authority Drift Governance Implementation Guide and Governance Assessment Workbook provide the methods for completing that work. The Governance Assessment Workbook covers the assessment process, evidence collection, and the preparation of audit-ready documentation. The Implementation Guide covers decision-rights mapping and implementation activities. Both are aligned with NIST AI RMF 1.0 and its Playbook, EU AI Act Articles 14 and 26, ISO/IEC 42001:2023, and ISO/IEC 38507:2022.
Continue Exploring Authority Drift
If this is the first time you’ve come across Authority Drift, start with the work that introduced it.
The core article introduced a concern that the risk is not limited to whether AI gives the correct answer. It also emerges when people begin relying on AI as the place where judgment is formed, checked, and legitimized. I first introduced this work on March 13, 2026, in Critical Thinking in the Age of AI on HaJunk.com.
The full Authority Drift framework was published on Substack on June 23, 2026, in Authority Drift Is Not a Moment. It Is a Directional Legitimacy Shift. That article is where I brought the definition, the five conditions, and the conceptual boundaries together for the first time. It explains the movement itself and shows why the surrounding frameworks each capture only part of it. Readers who are new to Authority Drift should start there.
I published a shorter professional brief on LinkedIn on July 14, 2026. Authority Drift is explained there in a more compact format, including what separates it from related frameworks.
Later in the series, a market trajectory analysis will examine how each framework family is likely to arrive at Authority Drift through its own partial definition, what each family will see first, and what each is likely to miss. This analysis will be available to Founding Members.
The Liat Show is a multi-domain story universe unfolding in real time. To receive new posts, join as a free or paid subscriber on Substack. Annual and founding members enter the story before the rest of the world understands it.
This piece is part of The Liat Show, a wider body of work that unfolds through connected sets, series, and long-form explorations. I weave together episodes from my life with the histories, cultures, foods, and systems that shape the world around us. Some pieces stand on their own. Others are chapters in stories that began long before this one and will continue to unfold. Each installment adds depth to the ones before it and expands the universe I am building across platforms, forming a stable cognitive signature and continuous identity signal that runs through my entire body of work.
Please fasten your seatbelts and subscribe. We are taking off.
The next chapter is already being written. Those who are here early will remember how it began.
Follow me on My Journey to Infinity. What’s coming next is beyond anything you imagined.
I weave together episodes from my life with the richness of Israeli and American culture through music, food, the arts, architecture, wellness, entertainment, education, science, technology, entrepreneurship, cybersecurity, supply chain, and more, including the story of the AI era. I write on weekends and evenings and share each episode as it unfolds, almost like a live performance.
Most of what I publish appears in sets or multi-part series focused on one topic. Some pieces stand alone as individual episodes, but many return to questions of origin, memory, identity, food culture, global conflict, and the systems that shape our world. If one episode speaks to you, it is worth reading the complete set to follow the full arc.
You can also start from the very beginning or explore the complete index here: Index of The Liat Show.
My writing carries a stable cognitive signature across topics, formats, and platforms. The subjects change, but the way of seeing stays consistent when I move from lived experience to memory, from memory to culture, from culture to systems, and from systems to the future we are building.
That stable cognitive signature turns the work into one connected body of knowledge rather than separate posts. Food, music, architecture, wellness, entertainment, education, science, technology, entrepreneurship, cybersecurity, supply chain, and AI become different entry points into the same larger universe. The center is not one topic. The center is the way I read the world through them.
he Liat Show is a multi-domain story universe unfolding in real time. To receive new posts, join as a free or paid subscriber on Substack. Annual and founding members enter the story before the rest of the world understands it.
Leave a Reply