Google’s AI opt-out is now a Top Stories opt-out

Google's rolling out a Search Console toggle that lets you pull your content out of AI Overviews, AI Mode, and Discover's AI features. It's the setting UK regulators forced onto the table, and it's landing in US accounts now too. On paper, it's the control publishers have been asking for since AI Overviews launched.

Then NewzDash's tracking data landed this week and complicated everything. Top Stories carousels — the traditional news feature publishers actively want to appear in — are now rendering *inside* AI Overviews on US trending queries. Which means the opt-out isn't a clean opt-out. It's a bundled opt-out. Pull yourself out of AI, and you may pull yourself out of Top Stories too.

That's the loop. And Google built it.

The toggle does exactly what it says. That's the problem.

The Search Console setting is genuinely narrow. It removes your content from AI Overviews, AI Mode, and Discover's generative features. It doesn't affect regular Search ranking. It doesn't touch AI training (that's Google-Extended, a different control). It doesn't override Merchant Center or Ads. Google says changes propagate within a couple of days.

The toggle is honest about what it controls. Google's SERP is not honest about what counts as AI.

Read the docs and it looks like a clean surgical control. One dial, one outcome.

The problem is Google's own product design has stopped respecting the boundary between "AI feature" and "regular result." When a Top Stories carousel renders inside an AI Overview, the toggle can only make one decision for both. And Google has decided that the AI surface owns everything nested inside it.

The toggle is honest about what it controls. Google's SERP is not honest about what counts as AI.

This is the CMA outcome nobody planned for

The UK's Competition and Markets Authority pushed Google into this. The June conduct requirement mandated that Google give publishers granular control over generative AI use. Google shipped the toggle. Regulator, meet product team.

But the CMA scoped the requirement to AI features specifically — because that was the boundary that existed when the consultation opened. Between the consultation and the shipped product, Google reorganised its results so that traditional features started living inside AI surfaces. The regulatory scope and the product reality no longer line up.

I don't think this was strategic. I think it's what happens when a regulator scopes an intervention against a product surface that keeps moving. But the effect is the same either way. A control that was meant to give publishers a real choice now forces them to make a much bigger one than the CMA had in mind.

Who this actually hurts

The publishers with the most to lose here are the ones who depend on Top Stories placement for trending news traffic. National publications, regional news, trade publications covering fast-moving beats. For them, Top Stories is a distribution channel that still reliably converts to clicks — one of the few in the current SERP.

Opting into AI Overviews for those publishers has been a genuinely hard call. AI Overviews summarise their content, cite them inconsistently, and cannibalise the click through to the article. Opting out was starting to look like a defensible position — better to be absent from a low-yield surface than present as an uncredited source.

That calculus just changed. Opting out now potentially means being absent from Top Stories on trending queries too. And Top Stories is the one AI-adjacent surface that still generates meaningful traffic for news.

The publishers who were most likely to opt out are exactly the ones who can least afford to.

The measurement problem got worse, not better

I wrote last week about the Search Console AI Overview reports being a metric trap — the data is thin, the attribution is confused, and enterprise teams are budgeting decisions around numbers that don't survive scrutiny.

This week's development makes it worse. If you opt out and lose Top Stories placement, you won't see that loss cleanly attributed anywhere. You'll see aggregate traffic drop. You'll see Top Stories impressions decline. You may or may not be able to reconstruct why. Google isn't going to send you a report saying "you lost 40% of your Top Stories placement because you toggled the AI opt-out."

The decision to opt out has to be made mostly blind — before you can see the second-order effects. And once you've made it, you have to wait weeks to see what actually happened, then work backwards to guess why.

What this actually means

If you're weighing this decision — and any UK publisher with meaningful Top Stories traffic should be weighing it — the calculus has three parts.

First, whether your content actually appears in AI Overviews at meaningful volume. Most sites don't. If you're not getting cited, the "loss" from opting out is theoretical, and the Top Stories risk becomes the dominant consideration. Check your Search Console AI reports before you touch the toggle.

Second, whether Top Stories is a real channel for you. For B2B service businesses, most e-commerce, and most SaaS, it isn't. The opt-out is genuinely clean for you. For news publishers, trade press, and anyone whose beat produces trending queries, it's a live risk that has to be measured.

Third, and this is the one nobody's talking about — whether Google will continue to nest traditional features inside AI surfaces. Because if the answer is yes across more feature types (Shopping, Local Pack, Video carousels), the toggle becomes a much bigger commitment than it currently looks. Today it's AI + Top Stories. In six months it could be AI + Top Stories + Shopping + Local. The bundling problem compounds.

The honest read

The Search Console toggle is a good thing. Publishers should have this control. The CMA was right to force the issue.

But the version of the control that shipped is narrower than the choice publishers now have to make. Google's product design has widened the gap between what the toggle technically does and what opting out actually costs. That gap is going to grow before it shrinks, because there's no product incentive for Google to unbundle features from AI surfaces.

The lesson for anyone thinking about AI-search governance more broadly: assume that any control you're offered is scoped to the product surface as it exists today, and assume the product surface will change to reduce the utility of the control. That's not cynicism. That's what's happened between January's announcement and this week's data.

The toggle is real. The choice it offers is smaller than it looks.