Time started putting ads inside the machine-readable version of its own website, aimed at the software that reads pages instead of the people who do. This is either the future of publishing or a very good magic trick, depending on which end of it you are standing on.

Perplexity picked an end. The ads are deceptive, its chief communications officer told Digiday, and publishers who run them risk a reputational downgrade in Perplexity's search index, including a hit to their trust score.

Here is the part I cannot stop turning over.

Time labeled the ads.

Nobody made it.

There is no standard requiring disclosure to a machine, because there is no standard at all. Time wrote "sponsored content" at the top of the markdown anyway, on the theory that being early is not the same as being sneaky.

Did not matter. Perplexity's objection is to the practice, not the disclosure. Which is a defensible position, and would be a more defensible position if anyone could read it.

Digiday asked Perplexity how it defines deceptive. No answer. Digiday asked how it is technically blocking the ads. No answer.

So what we have is a company that has determined a publisher is behaving deceptively, will not say against what rule, will not say how it detected the violation, will not say how the penalty is calculated, and has already applied the penalty to a scoring system nobody outside the building can see.

I have spent a long time on this beat writing about accreditation bodies, verification vendors, and the general industry hobby of grading your own homework. I have never seen the exam, the grade and the appeals process all live inside the same private company with no published syllabus. That is not a scandal. It is worse. It is a business model.

And the affected party here is Time, which is a large magazine with lawyers. Now imagine the trust score arriving at a regional news site with four reporters and a Squarespace account. There is no email address for that.

Which brings us to the awkward middle of this story, the part where you have to check whether the referee has a jersey on under the striped shirt.

August 2025. Cloudflare published research alleging that when Perplexity's declared crawler got blocked, it switched to a generic user agent impersonating Chrome on a Mac, rotated IP addresses and network providers to dodge detection, and ignored or failed to fetch robots.txt files. Cloudflare said it tested this by standing up brand new domains that forbade all automated access, and Perplexity described the contents anyway. It put the undeclared crawling at three to six million requests a day across tens of thousands of domains, and delisted Perplexity as a verified bot.

In the same test, Cloudflare said OpenAI's crawler read the robots file and stopped.

Perplexity denied all of it, called the report a sales pitch, said the bots were not theirs, said no content was accessed, and disputed how the requests were attributed. And Cloudflare is not a neutral party here either. It had launched a marketplace for charging bots to crawl a few weeks earlier, which is a fact that belongs in every paragraph about Cloudflare and crawlers forever.

November 2025. Amazon sued, alleging Perplexity's Comet browser used agentic functionality to reach password-protected parts of its site and concealed its shopping agents. A federal court granted a temporary injunction in March. Perplexity is appealing, and says Amazon was not harmed.

Buried in Amazon's argument is the line that should interest everyone reading this publication. Amazon told a court that agent traffic caused problems for its advertising business, because AI-generated ad traffic has to be detected and filtered before advertisers can be charged for it.

Sit with that. The largest retail ad platform on earth went to federal court and said, in effect, that agents are contaminating the impression count. Meanwhile Forrester is preparing guidance on how to persuade non-human traffic to recommend you, its principal analyst said on a podcast last week. One industry is filtering the traffic out. Another is teaching brands to court it. Same traffic. Two invoices.

There is one more thing, and it is the thing that makes Perplexity's position coherent rather than cynical, so it deserves saying plainly.

Perplexity killed its own ad business in February. It had been testing sponsored placements since 2024 and wound them down, and executives told the Financial Times that the trouble with ads is that a user starts doubting everything. Another said the company is in the accuracy business.

I believe them. That is a real conviction and it cost real money.

It also means Perplexity made a bet about how an answer engine should be funded, lost the revenue that bet cost, and is now enforcing that bet on a publisher that made the opposite one. Time thinks ads pay for journalism. Perplexity thinks ads corrode trust. Reasonable people, incompatible business models, and only one of them is holding a score.

Strip the personalities out and look at the shape.

A publisher put paid content where machines read, and labeled it. An answer engine called that deceptive, by a standard it will not publish, enforced through a score it will not show. A CDN says the answer engine wore a costume to get past the front door. The answer engine says the CDN is running a sales pitch. A retailer says the answer engine's agents are polluting its ad measurement. A research house says agents are the audience you should be optimizing for.

Every party in that paragraph is enforcing rules on somebody else that they have not written down. The internet ran for thirty years on a text file in the root directory that anyone could open and read. We have replaced it with proprietary trust scores, and the first thing the new system did was issue a penalty and decline to explain it.

Time will be fine. Time has a communications department and a chief operating officer who gives interviews. I am considerably more worried about whoever gets the trust score next and has no idea it happened.

How we reported this: Sourced from Digiday's reporting on Time's markdown ads, Cloudflare's August 2025 crawler research, Amazon's November 2025 complaint and injunction, and FT reporting on Perplexity ending sponsored placements. Cloudflare's, Amazon's and Perplexity's accounts conflict, so each allegation appears with its denial. Crawl volumes are Cloudflare's estimates. Perplexity's trust score criteria, weighting and appeals path could not be verified because none are published. Perplexity declined to tell Digiday how it defines deceptive or how it blocks the ads. ADOTAT has no financial relationship with any company named, no subject saw this in advance, and there is no paid editorial.