What fresh hell is this?
(See title.)
Time Magazine has a separate version of its website with ads only AI can see
AI webpage crawlers are now being served their very own ads that ordinary visitors never see, with one firm's boss openly describing a strategy to influence what chatbots say about brands. The next time you ask Claude about where to bank, its answer may have been influenced by this bot-targeted content. We only have one …
-
-
-
-
Thursday 6th August 2026 05:56 GMT Ancientbr IT
Re: Brilliant.
Influencing an LLM bypasses adblockers - when a chatbot appears to summarize information in your request for enlightenment, that summary isn't subject to the adblocker technology, unless you explicitly follow up on any link the chatbot supplies to you and *then* your local adblocker gets the chance to do its job.
Judging by the flow of reports describing how users don't check what the chatbot delivers to them (as apparent fact), whether the user is a kid in school or a lawyer presenting their case to a court, the opportunity to poison the information well is already well established.
Scary.
-
-
Thursday 6th August 2026 15:17 GMT Phil O'Sophical
Re: Brilliant.
selected lies made up by brand owners...
Or even better, selected lies made up by by competitors. Imagine the Macdonalds website presenting AI visitors with negative 'facts' about KFC, or Easyjet advertising carefully-selected statistics showing Ryanair in a bad light? The possibilities are endless.
-
-
-
Wednesday 5th August 2026 22:21 GMT DS999
So I used to say
That the demand for bandwidth was going to slow down, because it has been driven by human senses. We went from text to mp3 to less than SD video to HD video and now 4K video. That's about it for human senses, there is little value in going beyond 4K. Yes obviously the overwhelming majority of data moving around isn't being directly consumed by people, but in the end it was all driven by the needs of humans as the ultimate consumers of information.
But AIs creating their own ecosystem will see it rise without limit. All the SEO adslop Google was serving up hoping to catch our eye will be served to AIs, because if you ask an AI something and it tells you you aren't following a link where you can see an ad. So what could be better than fooling the AI into treating your ad as "facts" to provided to end user queries? Since AI training works via reinforcement learning, there is no limit to the number of ads that brands will want to place in their path, so their "facts" can be better reinforced than the "facts" provided by their competitors.
-
-
Thursday 6th August 2026 05:06 GMT jake
Re: Any lawyers reading this?
No, of course not. One cannot commit that kind of crime against an AI, not when its wrangler(s) is/are actively and intentionally setting out to collect any and all data for it's so-called "training".
To say nothing of the fact that no AI is a legal entity, and thus can not be charged with a crime.
-
-
Friday 7th August 2026 18:18 GMT doublelayer
Re: Any lawyers reading this?
That is technically true. However, the two steps are being done by different people, which gives both of their respective lawyers ways to argue that they're not responsible in addition to the existing get out of consequences free fallbacks that apply to most LLM things:
Ad in paper lawyers: Everything we said there is valid advertising copy. It's not our responsibility if a chatbot that we didn't write or tell people to use isn't capable of generating good results based on it. Fallback: You have no proof the LLM told the user to use this bank because of our text; you can't track a prompt to a specific bit of training data like that.
Chatbot company lawyers: Our LLM is intended to search the internet and summarize what it says. It's not our fault if the pages online have text designed to show up in those summaries. We didn't write that or do anything specifically to promote it. Fallback: We wrote in the terms of service that this was for entertainment only and warned people to check the accuracy of results, so the user who used it for financial advice and didn't read any further was not using it as directed.
-
-
-
-
Thursday 6th August 2026 06:12 GMT Bebu sa Ware
ads only AI can see
being on the dafter side of daft, I was wondering what on Earth could Time flog to an AI agent - a subscription ? Do those agents have an expense account, or money being a bit deficient in the pockets department.
Reading the article I realised Time was subverting the AI agent's training with porkies that humans might baulk at — given human gullibility they must be rather egregious lies.
Another variation on subliminal advertising. If you cannot sneak into people's brains directly, subvert their strap-on brains.
According to her friend Vincent Sheean, Dorothy Parker wasn't joking when she exclaimed "What fresh Hell can this be ?" I don't suppose we are.
-
Thursday 6th August 2026 15:34 GMT Anonymous Coward
It won't work
Unless the AI is going to surface the Time page in a banking search (which it won't if they've been cloaking that text from the search engines), it won't take in this text except possibly as training material which won't count for much. It won't be used to "ground" replies to specific queries. Prove me wrong.
-
Thursday 6th August 2026 22:49 GMT doublelayer
Re: It won't work
This iteration might not work very well, as it seems they are hoping to get into the training data, but they only have one step more to take to change that. I find it a bit interesting that they're doing what people have suggested online ad platforms do for some time, and the next step for them is also one of those. Those who object more to the privacy violation that modern web ads bring than to the advertising itself have long suggested that serving the ads from the same site that the content is coming from would help, and that's what this system is doing. If it takes the next step and embeds the advertisements on pages with related content, related enough that it will come up in LLM searches, that could get the ads seen and make it feasible to continually sell the ads to new buyers since it would use fresh data each time. LLM creators could do this much more efficiently and profitably if they felt like it, though. Not ethically, but I assume everyone involved gave up on ethics a while ago.
-
-
Thursday 6th August 2026 18:05 GMT David 132
I see Slashdot has picked up this story...
...and with the attention to detail and great editing for which that site is so justly famous, has headlined it "Times Magazine..."
I can't decide whether that implies AI news-regurgitators would do a better job, or that Slashdot itself is now using AI slopmachines to do humans' work.
