Home Daily News Roundup AIO? More like, “Ehh, I Owe You Nothing”; Passing A (Regulatory) Review

AIO? More like, “Ehh, I Owe You Nothing”; Passing A (Regulatory) Review

SHARE:
A comic depiction of Google's ad machine sucking money out of a publisher.

The AIO RIP

Google began showing generative AI search responses – known as AI Overviews (AIOs) – almost one year ago, and has maintained that they’re a boon for web traffic. 

Both intuitively and anecdotally, this is nonsense, and there’s data to back it up, Bloomberg reports. AIOs link out less often and send more traffic to platform or data licensing partners, such as Reddit, Tripadvisor and YouTube. 

In small group settings, Google employees have apologized to publishers and acknowledged that sites with niche, authentic expertise – the type of content sought out by the generative AI models – have lost between 80% and 90% of search traffic, with little recourse.

The travel category is particularly hard-hit by Google’s search changes, Bloomberg notes.

Travel blog The Planet D, for example, got 90% of its traffic from Google Search – until last year, that is, when Google launched AIOs and the traffic spigot turned off. Its founders say that AIOs sometimes so clearly parrot their writing style that certain answers even employ their favorite Canadian slang terms.

They stopped updating the blog in 2024, after launching in 2008, because ad revenue plummeted – and isn’t coming back. 

To rebuild the business, the pair are transferring to YouTube.

0 Stars

There’s more than one way to skin a cat – err, combat fake reviews. 

First, the UK’s Competition and Markets Authority (CMA) recently passed new measures to ban “outlandish fake reviews” and “sneaky hidden fees,” per the regulator’s press release

In addition to outlawing drip pricing practices, which tack on fees after a consumer has already begun the checkout process, the provisions also compel businesses to take “reasonable and proportionate steps” to remove fake reviews, reports The Verge

Presumably, the onus is on the businesses themselves because of how subjective reviews can be. One man’s five-star experience is another man’s one-star disaster, after all.

But that hasn’t stopped Google Maps, for example, from trying to automate the process with Gemini. According to Search Engine Journal, Google’s AI tool can now flag suspicious profile edits (such as switching business categories) and warns users when too many five-star reviews have been recently posted or removed.

Of course, even with these news tactics, fighting scams and fake business practices is a bit like playing Whac-A-Mole. But at least it’s better than nothing, right?

Fool Me Once

What’s this? Reports of Meta artificially boosting metrics to make its products look more palatable? Where have we heard that one before? 

According to TechCrunch, recent rumors from an unnamed whistleblower on Meta’s AI team suggest that the company inflated the benchmark scores of its Llama 4 Maverick and Scout models by training on “test sets.” 

Ahmad Al-Dahle, VP of generative AI at Meta, denied the rumors in an X post on Monday. However, he also copped to “reports of mixed quality across different services” since both products were released on Friday.

So far, the Meta Advantage+ ad suite has been mostly siloed from the rest of the platform’s consumer-facing AI products. But “Business AI,” Meta’s most recent offering targeted at small and midsize businesses, will almost certainly use Llama to create its agentic, chat-based customer support tools moving forward.

Which means that it’s worth keeping an eye on Llama’s development, even if – no, especially if – there are discrepancies in how the tool performs.

Because as we all learned when Facebook overinflated its video metrics almost a decade ago, these kinds of moves can have pretty disastrous consequences.

But Wait! There’s More

Investors are so desperate for good news that a fake tariff-related headline briefly caused the stock market to bounce back for a few hours on Monday. [Business Insider]

Trump threatens another 50% tariff increase on China, bringing the potential total to 104%. [Axios]

YouTube changed how it counts views for short-form videos, and creators aren’t sold. [Digiday

How Substack limits how publishers can monetize and redirects subscribers to its app rather than individual newsletters. [Journalists Pay Themselves newsletter]

Yes, video ads can show up in Google Search. [Search Engine Roundtable] And Google Search’s AI Mode can now search within images. [The Verge]

DOGE is planning a hackathon aimed at creating a “mega API” for accessing IRS taxpayer data. [Wired]

You’re Hired!

Moloco names Pat Copeland as general manager of its commerce media offering. [release]

Must Read

Critics Say The Trade Desk Is Forcing Kokai Adoption, But Apparently It’s Up To Agencies

Is TTD forcing agencies to adopt the new Kokai interface despite claims they can still use the interface of their choice? Here’s what we were able to find out.

Why Big Brand Price Increases Will Flatten Ad Budgets

Product prices and marketing budgets are flip sides of the same coin. But the phase-in effects of tariffs, combined with vicissitudes of global weather and commodity production, challenge that truism.

The IAB Tech Lab Isn’t Pulling Any Punches In The Fight Against AI Scraping

IAB Tech Lab CEO Anthony Katsur didn’t mince his words when declaring unauthorized generative AI scraping of publisher content “theft, full stop.”

Privacy! Commerce! Connected TV! Read all about it. Subscribe to AdExchanger Newsletters
Comic: Gamechanger (Google lost the DOJ's search antitrust case)

Here’s Who’s Testifying During The Remedy Phase Of Google’s Ad Tech Antitrust Trial

Last week, the DOJ and Google filed their respective witness lists and the exhibit lists for the remedy phase of the ad tech antitrust trial. Lots of familiar faces!

MX8 Labs Launches With A Plan To Speed Up The Survey-Based Research Biz

What’s the point of a market research survey that could take weeks, when consumer sentiment is rollercoasting up and down every day? That’s the problem MX8 Labs aims to tackle.