Dear Reader,

The news gets weirder every day. It isn't hard to find something weird to write about, but the problem is that weird is the new normal, so the same old dystopian topics are starting to sound monotonous. Not that we shouldn't learn about such things, but in this small and confined space, where I have but a page to tell you about something, it is hard to avoid those omnipresent themes that are resonating in the press. AI, in particular, does not seem to be a rock you can throw and be done with. Just as AI sucks up the world's RAM, it appears to be sucking up the world's news cycles. Flock cameras, data center bans, lab breakouts, perv glasses, AI slop – the same topics echo endlessly in blog sites, news sites, podcasts, and traditional media.

I made a real effort to find a corner of the AI debate that hasn't been over-crunched, and I hit upon a story that is still very much in the news but, I believe, is receiving less attention than it should be getting. Several news sources report on a story that seems to have started with a blog post by software developer Vincent Schmalbach regarding content fed to chatbots at the TIME Magazine website. Schmalbach writes, "TIME is now serving two different versions of its website. Humans get the magazine. AI crawlers get a stripped down markdown copy with ads baked in that no person will ever see." [1]

According to the post, the content of the page is organized so that certain AI agents receive an extra burst of markdown text, organized in the way that AI agents like to "learn," that is intended to feed hype about an advertiser into a chatbot. Google's own bots don't see this hidden information, so it doesn't affect the site's search rank. Everyone reading the story is thus secretly "voting" to build up the importance of the advertiser's product with Claude or ChatGPT without even knowing they are doing it.

Other outlets picked up on the story. The Register explored several other TIME Magazine pages and found the same effect [2]. So far, it isn't clear whether other commercial websites have adopted this technique, but if it works for one, you can bet the others will follow. The ad tech vendor Mobian appears to be behind this new approach. According to the Register story, Media Marketing Mag Digiday tracked down Mobian CEO Jonah Goodhart, and he seemed quite proud of this strategem, stating "When you influence ChatGPT, you're influencing potentially all of ChatGPT. If ChatGPT changes what it says about [a brand], it's massive and it's more than any one campaign could ever do…With a human you influence one person."

The cat-and-mouse game of building brands and faking out search engines thus reaches a ridiculous new level of complexity that is totally invisible to users. The Internet has operated for years under a paradigm where a vendor gets brand attention when a visitor goes to a website and sees an ad. Under this new paradigm, the eyeballs of the web visitor are irrelevant. What really happens is, when the vistor goes to the site, it adds gravitas to a secret little trapdoor that causes the web server to brainwash a chatbot, which, in turn, will brainwash other unsuspecting users elsewhere on the Internet. The next logical step would be for the content providers to start to prefer stories that maximize this chat-poisoning effect over other kinds of content, which would further degrade the state of web information.

Of course, one could say that this development is a mere blip in the history of chatbots – that they will eventually wise up and learn to counteract this slippery gambit. (Pause for the irony that we're waiting for these super-intelligent machines to "wise up.") But actually, there is no particular reason for the chatbots to do anything about this phenomenon unless they start losing business over it – this certainly doesn't seem like something that they would slow down to fix simply for moral reasons.

If taken to its dark conclusion, without government action or user rebellion, this dubious practice could lead to the strange reality where everything on the Internet is all just bait for AI agents, and the real "information" getting exchanged is invisible content bought with advertising dollars that has nothing to do with the content the bots are supposedly crawling the web to learn.

At that point, it would be fair to say that "AI slop" would refer to the whole Internet.

Joe Casad, Editor in Chief