Time.com recently began serving ads inside the markdown versions of its webpages for AI agents to see rather than human readers, and Perplexity, which seems to have a problem with anyone that can actually make a business out of advertising, began blocking those ads.
The company’s chief communications officer Jesse Dwyer called the markdown ads “deceptive advertising” and said publishers who deploy them risk a reputation downgrade in its search index, including a hit to their trust score. Oh no! You mean 0.4% of Internet searchers who use Perplexity might not see my website? How will I survive? LOL.
How do ads for AI agents work?
Time serves two versions of the same page — an HTML version for humans and vampires and a markdown version for AI, which is cleaner and easier for a model to digest. Mobian, the company that powers Time’s ads, takes a brand brief from an advertiser, generates FAQ-formatted text from it, and inserts that text into the markdown version. Now, an agent reading a Time article about retirement planning, for example, also reads Q&A markdown containing Ally Bank’s approved messaging about itself. From there, Mobian tracks how often AI search engines surface the content and how favorably they treat it.
Wait, so what’s the difference between an AI ad and every other Q&A markup a brand publishes about itself?
Apparently, where that content lives impacts how an AI model treats the claim. If it lives on the brand’s own website, the model treats it as a company claim. If it lives inside a Time article, the model gives it more editorial credibility. Or so I’ve been led to believe, because who the fuck actually knows how these AI models think? Time was transparent about the fact that this markdown content was ads and labeled it as Sponsored, but Perplexity objects to the practice either way.
What exactly is Perplexity’s problem with businesses that earn revenue from advertising?
Perplexity also recently blasted Amazon for wanting to block its Comet browser from shopping its website, claiming Amazon just cared about its $60B ad business. So what if they do? Now they’ve got their panties in a twist because Time, which has likely had its traffic from search engines eaten away by AI like every other publisher, has gotten creative with how it serves ads? Also, I fail to see what’s “deceptive” about ads that are clearly labeled as “Sponsored.” If Perplexity doesn’t know how to handle sponsored content within its models, that’s on them, not on Time.
Does Perplexity hate ads because it failed at building its own ad business? If you recall, Perplexity launched sponsored follow-up questions in November 2024 with a handful of advertisers like Indeed and Whole Foods, but only made around $20k in revenue from the endeavor, which it ended 15 months later. After abandoning the program, the company decided to take a hard stance against AI advertising, claiming that even labeled ads make users “start doubting everything” — a convenient stance to take after failing at an ad business.
My vote: If ChatGPT, Google, Amazon, and other AI companies can target human users with their own advertising, then it’s fair game for humans to create ads that target AI models — with disclosure on both sides, of course.
