Oscar’s New AI Rules: A Job-Protection Battle That Was Always Going to Fail
On May 1, 2026, the Academy of Motion Picture Arts and Sciences published its rule updates for the 99th Academy Awards. Three of them concern AI. Acting awards will only consider roles “credited in the film’s legal billing and demonstrably performed by humans with their consent.” Screenplay awards require scripts to be “human-authored.” And the Academy “reserves the right to request more information about the nature of the use and human authorship.”
Most coverage read this as “the Oscars are pushing back against AI.” I’m not going to judge whether the rule is good or bad for film as an art form. But as someone who has spent seventeen years in organizational design and run a fair number of layoff projects, I see something in this rule I have seen countless times inside companies: an organization takes on a accountability it has no capacity to fulfill, then uses that unenforceable rule to accomplish something entirely different from what it claims in public.
An Organization Should Clear Three Tests Before Taking On an Accountability
Before any organization writes a rule or takes on an accountability, it should answer three questions. Can you define the accountability clearly? Do you have the tools to enforce it? And if you can neither define it nor enforce it, what does this rule become as it runs?
The first two questions decide whether a rule can be fulfilled. The third decides what an unfulfillable rule does to the organization. The Oscar AI rule fails all three.
Test One: It Can’t Even Define “Human-Authored”
The screenplay award requires scripts to be “human-authored.” But the rule never defines what human-authored means.
I generate a first draft with AI, then rewrite it line by line. Does that count as human-authored? I write a detailed outline and character bios, then have AI fill in the dialogue. Does that count? I have AI produce twenty versions, pick one, and revise it. Does that count? The rule gives no answer.
What makes it worse is that the rule’s stance on AI is self-contradictory. On one hand, it assumes AI cannot be an author, which is why a fully AI-generated script is ineligible. On the other hand, it assumes that the moment a person uses AI, the script’s “human author” status becomes suspect and subject to review. These two positions undercut each other. If AI is just a tool, then the person using it is the full author, and the rule has no grounds to doubt him, just as no one doubts a script’s authorship because it was typed on a computer. If using AI dilutes “authorship,” then the Academy is conceding that AI in fact carried part of the creative work, which makes AI an author in functional terms, so on what basis does the Academy say it can’t be counted as one?
The Academy has to call AI “not an author” and “an author” at the same time for this rule to hold. A standard whose core concept contradicts itself cannot, in principle, be fulfilled. Test one: the rule fails on definition.
Test Two: Its Only Tool Is “I Reserve the Right to Ask”
Even if the definition problem could be solved, enforcement is a non-starter.
The only tool the Academy has given itself is that general clause: the right to request more information. This is not investigative power. It is not a verification mechanism. It is a “tool” that depends entirely on the honesty of the person filing. A screenwriter says, “AI just helped me organize some research, the script is mine,” and the Academy has no way to disprove him. A script is not like a performance. With a performance you can at least see whether what moves on screen is a human. A script is pure text. Its creation process is invisible, untraceable, unverifiable.
The Academy knows this. The phrase “reserves the right to request more information” is itself an admission: I have no capacity to verify on my own initiative, I can only ask a question after the fact. Test two: the rule fails on tooling.
Test Three: A Rule That Controls Nothing Is More Dangerous Than No Rule
Undefined and unenforceable, what does this rule become as it runs?
It becomes a rule for which no one can be held accountable for the outcome. Suppose a film wins Best Original Screenplay, and it later comes out that AI was used extensively. What happens next? The screenwriter says the core conception was his, AI only provided negligible assistance (”negligible,” of course, being a thoroughly subjective notion). The Academy says its rules required AI use to be disclosed, and he didn’t disclose it honestly. The judges say they only assess whether the script is good, not whether AI was used. No party can be held accountable.
But the real danger is not that no accountable party can be found. The real danger is that this rule controls nothing while making everyone believe everything is under control.
In organizational design terms, this is worse than having no rule at all. With no rule, people at least know no one is handling this, so they stay alert and look for their own way out. A rule that spins in place while looking authoritative makes everyone involved stop staying alert, stop finding their own way out, because they believe someone is already standing in front of them. Test three: the rule fails on the false sense of security it creates.
The Rule Has Two Real Functions, One External, One Internal
By this point the rule’s true function is clear. It was never meant to keep AI scripts from winning, because it can’t. But it does have two real functions.
Externally, it deflects liability. As long as the rule exists, the Academy can claim “we protected human creativity.” If something goes wrong, the Academy can push the accountability onto “the person who didn’t disclose honestly.” Take the credit when nothing goes wrong, pass the blame when it does.
Internally, it pacifies. This is the rule’s real engine. A large group of people in the film industry are afraid of being replaced by AI. They need a response, they need to see “the organization taking action.” This rule is that response. It does not need to be enforceable. It only needs to make that anxious group believe “the Academy is holding AI back for us.”
These two functions are two sides of the same thing. Precisely because it can deflect liability externally, it can promise protection internally, because it never intended, and never had the capacity, to deliver that protection, and it can still pass the blame if something goes wrong. A rule can both shift costs outward and hand out a sense of security inward, on one condition only: that it was never meant to be actually enforced.
So Who Ultimately Pays the Cost of This Failed Rule?
The cost doesn’t disappear just because no accountable party can be found. It is merely transferred.
The first to pay are the top creators. The truly irreplaceable actors and screenwriters were never afraid of AI in the first place. Their irreplaceability comes from within, the way Roberto Bolle still stands at the center of the stage at fifty, unafraid of robots, because a robot cannot imitate the unspeakable individual choices in every one of his muscles. These people should be winning on the strength of good work. Instead they now have to first prove they are “human enough,” and that proof can’t be adjudicated. The rule does not protect them. It drags them into a process they don’t need and that works against them.
The second to pay is the audience. In the theater the audience only sees the finished work. Whether a story moves them, whether a performance makes them hold their breath, has nothing to do with whether AI was used on the script. But this rule may keep a good script and a good performance out of contention because of “impure origins,” and may even affect whether they get financed and made at all. What the audience loses is the good work it could have seen.
The third to pay is the entire market, and the timing could not be worse. Global box office has stalled for two consecutive years. North American box office in 2025 was more than twenty percent below 2019. More tellingly, if you strip out ticket price increases and count actual admissions, the figure is down nearly forty percent from 2019. And the fact that this decline began before Covid shows that the audience leaving is not cyclical but structural, a long-term erosion in the appeal of the content itself. Mid-budget original films are disappearing in large numbers. The market increasingly leans on a handful of big-IP sequels. Audiences grew tired of formulaic storytelling long ago. At precisely this moment, a rule that narrows the sources of creation rather than widening them is, indeed, “ill-timed.”
And this is exactly what exposes the motive. If the Academy and the film industry were truly acting against the downturn, for the health of the market, what would the rational choice be? To embrace the gimmick. In today’s market, a film loudly marketed as “screenplay entirely written by AI” is itself a marketing flashpoint, and might well outsell the hundredth formula picture made under the rule’s protection. An industry genuinely driven by commerce, facing a downturn like this, would not ban the gimmick. It would use it.
They chose to ban it rather than use it. That tells you the original purpose of the rule was never the market and never creativity. It was the job-security anxiety of one internal group.
I have run many layoff projects. I have seen this maneuver many times, and the features are almost always the same. Facing an irreversible structural change, an internal group does not look for a way out. It manufactures a set of rules that look like “protection,” cannot actually be enforced, and shift the cost onto someone else. Its function is never to solve the problem. It is to delay the problem, to make the people staying put feel “the organization is protecting me.” Drop this AI rule into the cases I’ve handled, and nothing about it stands out.
“Delay” Never Saved Anyone’s Job
I am not against a transition period. An industry needs time to adapt to AI. That is entirely normal, and it should have it. What I am against is turning “transition” into “resistance,” and turning “adaptation” into “delay.”
In the layoff projects I’ve handled, this maneuver ends the same way almost every time. It never actually saves anyone’s job. It only delays. And the real cost of delay is spending the time that should have gone toward finding a Career Option B on a position that was never going to hold.
This Oscar rule transferred a cost it could never bear onto top creators, the audience, and the market, while keeping for itself a position from which it can both take credit and pass blame. But what it really takes away is from the practitioners who should have used this transition period to find their next path. What the rule hands them is not protection. It is the illusion that “we’re holding AI back for you,” and the greatest effect of that illusion is to make them stay put, at ease, and not go looking for a Career Option B.
When an organization speaks to its own people through a rule it cannot even enforce, what it is protecting is never the thing it claims to protect. At most, it buys the people unwilling to face reality a little more time to stay where they are.