What happened

On 9 November 2026, some jurors for the Independent Games Festival (IGF) published a statement saying they would not vote for games that use generative AI in “any capacity.” Soon after, the developer of the indie hit Big Walk published its own statement. It said it used AI “sparingly” in the “final stretch” of development to “help work through difficult” parts of the game.

I couldn’t load the full article, so I’m only going on the opening of The Verge’s report. I’ll stick to what it says and leave out what it doesn’t.

The claim behind the ban

The jurors’ position is simple. If AI touched the game at all, it doesn’t get a vote. The idea underneath is that any use of generative AI taints the work, so a clean line is easier to defend than a judgement call.

I understand the appeal. A line you can state in one sentence is easy to enforce. Nobody has to argue about how much AI is too much.

Testing it

I start with a plain question. What problem does the rule solve? The jurors want to protect work that people made, and that is a fair goal. The question is whether “any capacity” protects it.

Look at what the rule treats as equal. A studio that prompts a model to mass-produce art, writing and levels with no direction is treated the same as a team that used AI sparingly in the last stretch to get past a hard problem. Those are two very different situations. One has nobody steering. The other has a team that decided where AI could help, looked at what came back and kept responsibility for the result.

So the ban measures the wrong thing. It checks whether a tool was present, and ignores whether a person directed the work, checked it and decided what shipped.

Slop is a strategy problem

I’ve said this before. Slop comes from the strategy and how the tool is deployed, not from the tool itself. A knife is a kitchen tool or a weapon depending on who holds it, and banning knives from the kitchen doesn’t tell you who is cooking.

Slop is what you get when you expect AI to think on its own. You hand it no direction and no posture, and you ship whatever comes out. The result is generic, and people can tell. Increasingly the search engines and AI systems can tell too.

Volume without a frame is the real issue, and volume alone is fine. If a person set the direction, reviewed the output and made the final call, the output carries that person’s judgement. If nobody did, it carries nobody’s.

What this looks like for an operator

Take a clinic, a shop or a marketing team. Someone on staff will use AI to draft a reply, summarise a report or fix a stubborn problem. A blanket “no AI” rule won’t stop that. It will push the use out of sight, and then nobody is checking the output.

A rule that names an owner works better. For every piece of work that leaves your business, one named person directed it, checked it and decided it could go out. That person is responsible for it, whether AI helped or not. This is the same principle I teach: you direct, the AI produces, you check and decide. You are responsible for every output that leaves your hands.

That rule is also easier to enforce than a ban. You can ask who owned a piece of work and get a name. You can’t easily prove nobody used a tool.

Where this shows up in my own work

I use Claude for almost everything. The way I put it is that Claude did the building, the counting and the checking, and I made the decisions. The decisions are the part nobody else can supply. Context, posture and opinion come from me, and that is why high-volume content can still carry a specific view.

I also don’t think the absence of a tool is a mark of quality. Good judgement marks quality. A team using no AI can still ship careless work, and a team using AI can ship careful work.

What a business should do

Don’t write a blanket rule about the tool. Write a rule about the output.

  1. Name an owner for each type of work that goes out to customers, such as replies, pages, ads and reports.
  2. Define what “checked” means for that work. A fact check, a read against your brand, or a look at the numbers.
  3. Keep the commitments that cost money in your systems. Refund limits, discounts and delivery dates should be enforced by the platform, not left to anyone’s judgement, human or AI.
  4. Ask staff to say when AI was used, so the owner knows what to check.
  5. Review the output after it ships, and fix the brief when something is wrong.

If a festival or a client sets a stricter rule, follow it. But for your own business, the useful question is who owns the result.

Source: The Verge AI: The indie game community is in disarray over AI and Big Walk