For publishers who make their money through advertising, it’s difficult to see bots as anything other than a threat. They crawl their content, chop it up into machine readable code, and then serve up the information as context in AI summaries, usually without leaving so much as a dime in the tip jar. With bot activity surging as human referral traffic plummets, it’s no wonder that many publisher sites block as many bots as possible. The media world may be beginning to rethink that premise. Could it be that bots may not just be a threat to defend against but an audience to cater to? At least one publisher is taking that point of view: Time has begun to serve ads specifically meant for bots in an effort to monetize what is arguably every publication’s fastest-growing audience.
Here’s how it works, according to Digiday: Building on an effort to create machine-readable versions of its pages, Time is including what is essentially an advertiser-directed FAQ section, tailored to answer questions that readers ask in AI search engines about the brand or its products. Those FAQs are only visible to bots, and when they crawl the page, the content is—hopefully—repeated in some form in the summary presented to the person who made the query.
Time is reportedly selling one agent ad per machine-readable page and pricing it as premium inventory. Those requests can be counted when a crawler fetches the page, which translates an impression-like metric, if a human-free one.
The idea is innovative in its simplicity: By seeing the scrape as an ad impression by any other name, the publisher can sell its machine pages as inventory. But the advertiser is really buying a possibility: the chance that an AI system will retrieve the sponsored material and carry some of it into an answer. There’s no guaranteed placement or even a guarantee that the message will appear.
For the publisher, the model shifts the expectation of payment from the AI vendor, who probably wasn’t going to pay anyway, to the advertiser. In fact, getting scraped is a necessary part of the model—without it, the whole thing doesn’t work. No one likes ads in their experience, but if they need to be there, people want them clearly labeled so they know what’s content and what’s commercial. Time includes disclosures in the bot ads they’re serving, but that doesn’t guarantee they’ll survive the trip from the source page to the answer. Depending on the query—and how aligned the ad is with the content—the information might be mixed in with the answer.
That means disclosure has to function as data, not just a label. The model must recognize the material as commercial, preserve that status as it processes the page, and tell the user when it affects the answer. There is no settled standard requiring AI systems to do any of those things.
Time is working with Mobian, an ad-tech platform, on its ads for bots, but there are other companies with versions of the same idea. One of them is Oasy, whose approach is more aggressive: its tech injects an advertiser message specifically intended for the bot and invisible to the human reader. I’ve tested Oasy’s publisher software, and it gives publishers a clear count of bot requests and ad impressions.
This is one important way publisher-side bot ads differ from the ads that ChatGPT and Google are serving in their AI experiences. Both emphasize clear labeling and separation of the ad from the answer. There’s more than a hint of irony here that the approach of big tech to AI ads has clearer separation between editorial and commercial than the one from the publishing world, but you can see why it’s turned out this way: Pay-per-crawl and pay-per-use models haven’t yet generated meaningful revenue for the industry, and licensing agreements are generally only available to big outlets. Publishers need to get creative with their monetization methods, at least when it comes to bots.
But the larger distinction—and the one most relevant to advertisers—is the value each model offers. If you choose to advertise in ChatGPT, you get clearly marked ads with guaranteed placement on a platform that has massive reach. A publisher offers something different: authority. If its content is broadly deemed authoritative by AI systems, that authority can travel across engines, including ChatGPT, Gemini, Claude, AI Overviews, Perplexity, and everyone else. While different AI engines weigh things differently (and licensing deals matter too), the value of an authoritative post, author or publication is potentially amplified in an AI ecosystem, even when its human audience might be comparatively small.
On top of that, the bot layer is even bigger than you might think. As agentic use cases increase, agents often spin up subagents to research and return information, seeking important context on the web. Similarly, in any AI search, there are many “fan-out queries”—searches on related topics that a human never sees. In other words, there will be many instances where bots themselves are the audience for these queries. The human may still be the ultimate reader, but the bot is the intermediary and may make the first cut.
In those cases, will the disclosure be carried through accurately? And even if it is, might the bot deem the information relevant to the query anyway? How will it communicate that a particular part of the answer is commercial in nature? What if the goal of the user is action, not information? When an agentic system needs to choose a tool or product to complete a task, could an ad influence that choice before the human even sees the options?
If the ethics are gray, so is the path back to the advertiser. An AI system may retrieve, paraphrase, omit or blend the sponsored text, leaving the advertiser without a stable creative unit, guaranteed placement or reliable link to measure. The industry would be selling influence over a recommendation without necessarily being able to show exactly how that influence appeared. On top of that, we don’t know how the AI companies will view this model. They might see it as innovative, and if it ends up taking off, it could even help ease the pressure from the media industry to pay for content. On the other hand, ad-supported platforms like Google and ChatGPT may see it as a competitive threat and train their systems to filter or downrank content with these kinds of ads. They could also treat bot-only promotional copy as cloaking, spam or an attempt to manipulate retrieval.
Credit to Time for testing a new kind of advertising, one that moves past the unrealistic expectation that AI systems will somehow start paying for the content they harvest. But the model depends on a delicate trade. Publishers are monetizing the authority that makes their information valuable to AI systems in the first place. If the advertising weakens that authority, the inventory could lose its value.
While the media industry has experience striking that balance, in this case generative systems are a wild card. Ultimately, publishers don’t control how AI systems work, or what the companies who build them will do. Ads for bots are a promising shot at real revenue—provided the systems that make them valuable don’t move the goal posts.