Colossal is one of the most-read art publications on the web and a source we cite here most weeks, including for the monthly open-calls roundup that this site’s Community coverage depends on. On September 16, editorial director Grace Ebert published a letter setting out where the magazine stands on AI: everything stays human-made.
What they committed to
The commitment is specific rather than gestural. It covers “selecting images to writing headlines to compiling our daily newsletter” — that is, not only the writing, but the editorial judgment calls that are exactly where automation is currently being sold hardest. No algorithmic decision-making in the process.
The context makes it a real commitment rather than a cheap one. It’s a four-person team, publishing daily, funded by memberships. This is precisely the size of operation that the tooling is pitched at: small staff, high volume, thin margins, obvious efficiency gains available.
The letter’s argument is that art is a uniquely human capability that drives empathy, that AI threatens both livelihoods and “the very substance of our humanity,” and that creating the world you want to live in means embracing messy, idiosyncratic human creativity.
Why the friction line is the interesting part
Most editorial AI statements are defensive — a policy page saying we may use AI assistance with human oversight, which is broad enough to mean anything. Colossal’s is unusual in that it names what the choice costs and keeps it anyway: the team deliberately maintains friction, and describes the human-powered process as resource-intensive.
That’s a more honest position than the alternatives on offer, and it’s worth separating from a claim it doesn’t make. Colossal isn’t arguing nobody should use these tools; it’s a statement about what this publication is, in the same category as a gallery saying what it will and won’t show.
The awkward part, stated plainly
This site’s position is different, and it would be dishonest to run a piece about Colossal’s stance without saying so. NewsNix covers AI tools as a core beat, publishes tutorials on generative image editing, and reports on models weekly — and this publication is not produced the way Colossal’s is. So take what follows as interested commentary, not a neutral verdict.
The part of Colossal’s argument that survives regardless of where you land is the one about judgment: what an arts publication is worth is the choosing — which work gets attention, which claim gets checked, which press release doesn’t get repeated. That’s the thing a reader is actually trusting, whoever or whatever assembles the page.
Where their letter lands hardest is on the point about small teams. The efficiency case for automating editorial work is strongest exactly where the editorial voice is most distinctive — small, opinionated, independent outlets. If those outlets automate, the web loses the thing that made them worth reading, and it happens quietly, one headline generator at a time.
The counter-case is that the same tools let a four-person team survive at all. Colossal has looked at that trade and declined it. That’s a data point worth having, from a publication with the readership to make it consequential.