Buying ads automatically used to feel like magic. The big promise of “programmatic advertising” was simple: brands could instantly match their ads with the right shoppers across millions of random websites.

But things have changed. In 2026, the open web has turned into a massive trap filled with fake pages.

Because it is so incredibly easy to use generative AI tools like ChatGPT, scammers have built a huge number of “Made for Advertising” (MFA) websites. These are cheap websites packed with useless AI-written articles. Their only purpose is to load as many ads as possible and steal money from brand advertising budgets.

Marketers are finally waking up to a harsh truth. If you buy ads across the open internet without checking exactly where your money goes, a massive chunk of your budget is just paying fake AI bot farms.

How the AI junk site problem got so big

Made for Advertising sites have actually been around for years. But AI completely changed the game.

In the past, building a network of clickbait sites was hard work. You had to hire cheap writers to steal and rewrite articles all day. Today, a scammer can launch 50 fully active websites over the weekend using self-driving AI bots. It costs them basically nothing.

These sites are built purely to trick ad systems. They cram too many ads onto the screen, the pages refresh on their own to load new ads, and the actual articles make no sense to real humans. But because the content costs nothing to make, these fake sites are grabbing a huge share of the ad market.

Industry watchers estimate that up to 30% of a brand’s open-web budget can quietly disappear into these fake AI sites. Brands are effectively paying thousands of dollars to show the same high-quality video ads to simple bots that just pretend to be humans scrolling.

Why standard safety tools aren’t catching the fake ads

For years, brands leaned heavily on software tools to keep their ads safe. These “brand safety” tools scan websites and automatically block ads from showing next to hate speech, violence, or highly toxic political news.

But these automated safety tools are totally failing to catch AI-generated filler articles.

The problem is that the text on a fake AI site isn’t angry or offensive. It’s just boring and useless. When standard software scans the page, it sees perfectly safe, well-spelled words. So, the software happily approves the ad.

This shows a much deeper problem with how ads are bought today. The whole industry relies on computer code checking other computer code. When brands hand over total control of their budgets to automated platforms—as we explained in our guide on how black box advertising hurts brands—they lose the ability to see where their ads are actually running until all the money is gone.

The lack of basic transparency means a brand might celebrate getting incredibly cheap clicks on a Tuesday, only to find out on Friday that every single click came from a fake, robotic website.

Smart buyers are going back to direct deals

The days of buying ads blindly across millions of unknown websites are ending. To fight the flood of AI-generated garbage, smart media buyers are choosing quality over sheer size. They are walking away from the “open web.”

• Instead of trying to keep a massive, never-ending list of bad websites to block, smart agencies are creating strict “allowlists.” These lists explicitly name the few hundred verified, real-human publications where their ads are allowed to show.
• Media planners are skipping the risky open auctions entirely. Instead, they are signing guaranteed deals directly with premium publishers (like top news sites and magazines) to make sure real people see their campaigns.
• Brands are finally stopping measuring success just by “viewability” (whether the ad loaded on a screen). They are starting to track real engagement, like how long a user actually reads the page, which fake AI sites cannot fake.

This defensive move is a massive headache for the people buying the ads. They can no longer just type a budget into a dashboard and let Google optimize for the cheapest click. As we pointed out in our look at how ad agencies are surviving the AI wave, the main job of an ad buyer has changed. Their job is no longer pushing buttons; their job is heavily auditing the supply chain to make sure the inventory isn’t fake.

What this means for your budget next quarter

Moving away from the open web means you have to completely reset how you measure success. If your media agency is bragging about getting insanely cheap ad impressions on your display campaigns right now, you need to demand a report showing every single website address where the ads ran. In 2026, if an ad is shockingly cheap, it is almost certainly running on a fake AI site.

The friction of implementing a strict allowlist:

When you change your buying strategy from the broad open web to a strictly verified “allowlist,” your dashboard metrics will break instantly.

Because you are suddenly cutting out millions of hyper-cheap, zero-value bot clicks, your overall Cost-Per-Acquisition (CPA) will frequently look two to three times more expensive overnight. The numbers will look terrible at first glance.

You must proactively warn your finance department and your Chief Marketing Officer before you make this switch. You need to explain in plain English that your marketing isn’t suddenly performing worse. You are simply starting to pay fair market prices for real human attention, instead of quietly funding an AI bot farm.

If you don’t reset these expectations with your bosses before making the switch, you will face massive pressure to turn the cheap, fake traffic back on just to make the monthly spreadsheet look green again.