They're Turning Your Content Into AI
And you’re the one who’s gonna get blamed for it.
There was a dust-up on Sunday where a bunch of people got mad at Bookshop.org for a Meta ad of theirs that appeared to be AI. But Bookshop.org doesn’t seem like the sort of place that would do that, so @overflowingshelf investigated:
Guys, before you attack @bookshop_org over the AI ads, THIS is the creative they uploaded (from the Meta ads library if you want to check for yourselves.) Meta is forcing AI into their business suite and is making it harder and harder to turn it off as they want AI to run the WHOLE ad process - from copy to creative to audience building. It is a nightmare for anyone who works in advertising. (Threads)
It appears that Bookshop.org didn’t use AI; Meta did, by automatically creating and serving at least one AI-made variation of the original, non-AI content.
Down-thread, @thehawthornelibrary posted a walkthrough on how to turn off Meta’s AI ad features. But Meta often silently re-enables features after updates, so you have to keep checking your ad setups. And they’re constantly adding new features, and changing and reorganizing things, and A/B testing different configurations with different groups of users, so any guidance is likely to be outdated by tomorrow and may not even apply if you’re in a different, secret testing group. And you don’t even get any kind of review or notification before your content is AI-ified and served to the public under your name.
I went in to our Shiraki Press ad account to ensure the offending features were disabled. Perhaps predictably, some were not. But when I clicked to turn them off, I was served a popup demanding a reason—I had to pick from a drop-down—and the highlighted confirmation button was “leave this feature on”.
This is all so fucking hostile.
These giant tech companies are convinced they know our needs better than we do. To wit, here is Zuckerberg himself on their future vision for advertising on Meta (Stratechery):
“But basically, we believe at this point that we are just better at finding the people who are going to resonate with your product than you are… you come to us, you tell us what your objective is, you connect to your bank account, you don’t need any creative, you don’t need any targeting demographic, you don’t need any measurement, except to be able to read the results that we spit out.”
He’s confidently wrong. We’ve experimented with several ad campaigns this year, trying out different configurations—to the extent that the platform allows us to configure things, which is increasingly “very little” as they continue to drive everyone into their automated Advantage+ program—and Meta has consistently mis-targeted our ads, primarily to retired white men. Meanwhile, our direct experience at conventions, markets, and book stores is that Gen X, Millennial, and Gen Z women and genderqueer folks are by far our most enthusiastic and ready-to-buy audience. But since Meta has been very poor about placing our ads in front of those demographics, those ads don’t convert to sales, and Advantage+ gives us no “advantage” at all.
But of course they don’t care about the AI being right, they just care about it being AI. And if we’re just gonna AI everything, then what’s the point of doing anything any more? They’re selling us a future in which we don’t think, we don’t solve, we don’t create, we just sell. Sell what? Who cares? Fuck you, the line’s going up! Meanwhile, data centers filled with non-auditable code1 determine who gets rich and who dies poor, and their owners call it a meritocracy.
For now, we’ve pulled our ads from Meta. It feels scary to give up on advertising to half the world’s population, but 1) I sincerely question whether that figure is even verifiably true, and 2) as I noted above, they’re not serving our ads to the vast majority of those people anyway.
The hellscape makes us feel like we have to be on these platforms, like we have to let them siphon our attention, our energy, and our money, like we have to play their games for our businesses to survive… but it’s all just economic abuse perpetrated by TESCREAL sociopaths. We can choose a different way.
-
“The technical challenges of AI auditing are rooted in the inherent attributes of the AI workflow and are central to its functioning. A significant portion of these challenges are due to the accelerated and continual increase in complexity and opacity of "black box" AI models, including Generative AI models trained on massive datasets. The complexity of these models correlates with the opacity of their underlying mechanisms; in other words, the lack of explainability of black-box models poses significant barriers to AI auditing (Thelisson, 2017; Knowles and Richards, 2021; Herrera-Poyatos et al., 2025; Pégny et al, 2019; Peltovic, 2023).” (arXiv) ↩
You just read issue #4 of The Hellscape. You can also browse the full archives of this newsletter.
Add a comment: