Reuters and Politico are among the publishers reconsidering a bargain the web has taken for granted for two decades. The Wall Street Journal reported on 22 July 2026 that major publishers were weighing whether to stay in Google Search at all. The reason is structural, not emotional. Every control that reaches AI Overviews carries a cost. None of them leaves a publisher indexed, snippeted, and out of AI Overviews at once.
The Case for Blocking Google Search
The trade used to be simple. Publishers let Googlebot index everything, and Google sent readers back. AI Overviews changed what the index is used for without changing how a publisher can respond to it.
Mike Reed, CEO of USA Today Inc., described the logic his company already applies to crawlers without a licensing deal:
“I wouldn’t call it a big decision because we’re blocking other crawlers. For those with licensing agreements, they get our content. For those without, we block them.”
— Mike Reed, CEO, USA Today Inc.
Reed was describing crawlers without licensing terms, not Google itself. But it is the same arithmetic publishers say they would apply to Google. If access stops paying its way, it gets cut off.
Reuters is weighing blocking Google’s bot for its consumer-facing news product, per President Paul Bascobert. Politico staff have discussed limiting Google’s and other bots’ access to free articles, as Adweek reported.
The Traffic Numbers Behind the Threat
The urgency has a data trail. Similarweb data puts Newsweek’s own visits at roughly 100 million in May 2025, falling to about 23 million by June 2026. That is a drop of around 75%. Semrush found some publications declined more than 40% from June 2025 to June 2026.
Not every outlet lost ground. The Guardian and the BBC both gained traffic in the same window, per Semrush. That matters. Whatever is driving the decline elsewhere, it is not a uniform tax on every publisher.
Reddit Is Running the Same Math, With a Number Attached
Reddit shows what happens when this arithmetic gets a price tag. The Journal reported that renewal talks over its AI data-licensing deal with Google have hit turbulence. The original deal, signed in February 2024, was worth roughly $60 million a year.
If that fee stops covering what the access is worth, non-renewal would be a spreadsheet decision rather than a protest.
Reddit’s stock fell the same day. RDDT closed 22 July at $170.38, down $15.46, or 8.32%, wiping out close to $3 billion in a single session. A Reddit spokesperson told CNBC the company negotiates “just like any business should, by focusing on doing what’s best for Reddit.” The spokesperson added, “A lot has changed since those first deals were signed.”
What Controls Actually Exist Today
Before any publisher pulls the trigger, it is worth being precise about the levers. Google’s own robots meta tag documentation spells out how the snippet directives behave. Google-Extended and a robots.txt block sit outside that page, but they belong in the same comparison. None of the five lets a publisher stay indexed, keep its snippet, and keep its content out of AI Overviews and AI Mode.
Site owners can generate the crawler directives for their own domain. The trade-offs do not change based on how the file is written.
| Control | What it blocks | Blocks your content in AI Overviews? | What it costs you |
|---|---|---|---|
| Google-Extended | AI training use, per Google’s own description | Not established. Google frames it as training-only, and points to a separate Search Console control still in testing | Nothing on Search visibility |
| nosnippet | Snippet and video preview, plus use as a direct input for AI Overviews and AI Mode | Yes, as a direct input | No text snippet in any Google result. A static thumbnail may remain |
| max-snippet:[number] | Caps snippet length, and limits how much can feed AI Overviews and AI Mode | Limits, not a full block | Snippet truncated everywhere to the count you set |
| data-nosnippet | Chosen text spans in span, div and section elements | Not established | Google says not to add or remove it on existing nodes with JavaScript |
| robots.txt disallow Googlebot | Crawling by Googlebot | Not established | Google stops crawling. Note robots.txt governs crawling, not indexing; the documented removal control is noindex |
Can you block Google’s AI without losing search traffic?
Not cleanly. Google-Extended opts you out of AI training while leaving Search visibility untouched, but Google does not present it as a control over AI Overviews. The only documented levers that reach AI Overviews are nosnippet and max-snippet, and both cost you your ordinary search snippet.
The Pressure Building Behind the Scenes
Cloudflare has been public about wanting a middle path. Chief Strategy Officer Stephanie Cohen said the company wants “a technical solution that allows you to be discoverable without having to give your content away for free.” It is the same logic behind treating crawler access as a measurable give-get channel, which Cloudflare and beehiiv built for newsletter publishers instead of an all-or-nothing switch.
Google has acknowledged the gap. On 28 January 2026 it said it was exploring new controls. Sites would be able to opt out of AI-powered search features without losing traditional search placement. That came amid pressure from the UK’s Competition and Markets Authority. It has not shipped.
A Google spokesperson points to the company’s current and in-testing controls, describing “clear, granular controls to manage their content, including Google-Extended for AI training and a new Search Console control we are testing for generative AI Search features, neither of which impact traditional Search visibility.”
What Publishers Are Actually Protecting
Future’s Chief Revenue Officer Mike Peralta put the underlying stakes bluntly:
“AI is nothing but a hallucination without our content—not just Future’s content, but every publisher’s.”
— Mike Peralta, Chief Revenue Officer, Future
That is the leverage argument underneath every threat to block Google. Content is the input AI systems need, and publishers are testing how much that buys them. For now the available tools are blunt. Until Google ships the granular opt-out it says it is testing, pulling out of Google Search stays a last resort rather than a routine move.