Your AI Committee Is Not Slow. It Is Answering the Wrong Question.

The US government shut down Claude Fable 5 on a Friday afternoon and turned it back on nineteen days later. Your AI committee meets again in October.

That is not a joke at the committee’s expense. It is the whole problem in two sentences. On June 12, 2026, at 5:21 PM Eastern, Anthropic received an export control directive suspending all access to Fable 5 and Mythos 5 for any foreign national, inside or outside the United States. The company disabled both models for every customer. On June 30 the Department of Commerce lifted the order and access came back the next day. A frontier model appeared, vanished, and returned inside a single summer month.

Now count how many times your academic technology committee met during those nineteen days.

The number everyone quotes, and the number that actually matters

EDUCAUSE surveyed 1,960 higher education staff and faculty in the autumn and published the results this year. Ninety four percent had used an AI tool for work in the previous six months. That is the number that gets quoted in the keynote.

The number that matters is the second one. Forty six percent of those same respondents were unaware of any institutional policy guiding work related AI use. Fifty six percent had used AI tools their institution did not provide, which means tools nobody had checked for data privacy, accessibility, or intellectual property exposure.

Read those together and the usual conclusion is that people are not following the rules. That conclusion is comfortable, and it is wrong. Almost half of a workforce does not accidentally miss a policy. If forty six percent cannot find the rule, the rule is not where the work is happening.

The three clock problem

Writing in Times Higher Education this month, Nick McIntosh, a learning futurist at RMIT University Vietnam, gave this its proper name. Universities are running three clocks at once. AI development moves in weeks. Institutional governance moves in semesters. Practitioners live in the gap between the two, every single day.

Nobody set out to build that. The semester clock is not laziness. It is the correct speed for the decisions a university normally makes, because most of those decisions are meant to hold. A degree requirement should not change between October and November. A grading appeal process should be the same in March as it was in September. Deliberation at that pace is a feature.

Then a tool arrived that does not produce permanent decisions, and we handed it to the body built to make them.

Why a faster committee does not fix this

The instinct is to speed the committee up. Meet monthly. Add a standing item. Create a rapid response subgroup. Every one of those is an attempt to make a semester clock tick like a weekly one, and it fails for a reason that has nothing to do with effort.

A committee produces a decision. A decision is a thing you make once and then defend. That is exactly the wrong output for a tool whose capabilities, price, vendor terms, and legal status can all change before the minutes are approved. Ask a decision making body to govern a moving target and you get one of two outcomes. It decides too slowly to be useful, or it decides quickly and is wrong in public.

There is precedent for what happens when an institution wins the technology and loses the governance. We did this with social media. The platforms were built here, scaled here, and shaped an entire generation’s attention here, while the governing conversation ran a decade behind. The bill arrived anyway. It just arrived as a cost nobody had budgeted for.

What governing a fast tool actually looks like

Every field that already governs something faster than its own committee cycle solved this the same way. It stopped producing decisions and started producing a loop. The shape is consistent enough to name.

You log what is actually being used. Not what was approved. What is running. Fifty six percent of that EDUCAUSE sample is invisible to their institutions right now, and you cannot govern what you cannot see.

You make every position reversible. A rule you can roll back in a week is a rule you can afford to set this week. Permanence is what makes a body slow, because permanence is what makes being wrong expensive.

You build a review queue for the exceptions. The committee’s job stops being to anticipate every case and starts being to handle the cases that actually surfaced. That is a smaller job, it arrives with evidence attached, and it fits inside a semester.

McIntosh’s own recommendation lands in the same place from a different direction. Name someone, a role or a small team, and give them a standing channel into governance. Not an annual survey. A live line from where the work happens to where the rules get written.

The part you control

Here is the honest constraint. You are probably not going to rebuild your institution’s governance calendar this semester. That is a real limit and pretending otherwise wastes your time.

But the three clock problem is fractal. It repeats at every scale, including yours. Your course has the same gap: a syllabus written in August, a tool that changed in September, and students making decisions in the space between. At the course level you are not a committee. You are one person who can write a rule on Monday and revise it on Friday, which means you already have the thing your institution lacks.

So govern the scale you actually own. Write the AI clause for one course, with a date on it. Keep a running log of what you allowed, what you refused, and what you changed your mind about. Put the rule inside the assignment where the student meets it, not in a policy portal they will never open. That is the whole method, and it is why forty six percent cannot find their institution’s policy: it was never written where the work was.

This is what SeedStacking means in practice. Seed is the first written clause. Sprout is the log that catches what the clause missed. Grow is the revision you make with evidence in hand. Harvest is a Course Record that carries your actual reasoning into next term instead of starting from a blank page again. A decision you made once is a rule. A decision you keep dated and revisable becomes Teaching DNA.

The seed

Your committee is not slow. It is being asked to make permanent decisions about something that refuses to hold still, and it is doing that job about as well as anyone could.

The question was never how fast can we decide. It is what do we build so that being wrong in October is survivable in November.

Nineteen days is how long a frontier model can disappear. Your syllabus has to survive that, and it is the one document in this entire argument that you can rewrite by yourself this afternoon.