Full disclosure: This post contains affiliate links. If you make a purchase through these links, 99signals may earn a commission at no additional cost to you.
For about two years, the dominant content playbook came down to three words: publish more, faster, cheaper.
AI made it possible to spin up hundreds of articles a week, automate entire publishing pipelines, and flood LinkedIn with polished posts that sounded professional while saying nothing.
That playbook just hit a wall, and it hit on two fronts at the same time.
On May 21, 2026, Google started rolling out its second core update of the year. It finished on June 2, after twelve days of ranking volatility that hit scaled AI content sites hard.
In the same window, LinkedIn confirmed it was deploying AI classifiers to suppress generic AI-generated posts, claiming 94% detection accuracy in early tests.
So the same underlying problem surfaced across search and social within a few weeks of each other. If your strategy leaned on AI volume for both channels, you’re now exposed on both.
Table of Contents
What Happened on the Search Side
Google’s May 2026 core update was officially described as a routine update to surface relevant, satisfying content. That’s the standard language Google uses for every core update.
But the timing and the pattern tell a more specific story.
This update landed 48 hours after Google I/O 2026, where Google confirmed AI Mode had passed one billion monthly users and deployed Gemini 3.5 Flash as the default AI Mode model globally.
The core update and the AI search overhaul are running on parallel tracks, and they compound each other.
Here’s what practitioners reported during the rollout: sites built on scaled AI content, programmatic publishing pipelines, and automated content agents saw sharp visibility drops within days. Not weeks. Days.
The pattern isn’t new. Every core update since the Helpful Content Update has pushed in the same direction: original content, expert voices, useful answers, less thin SEO bait.
What’s different this time is the severity and the speed. When AI lowered the cost of producing mediocre content to near zero, the volume of mediocre content exploded. And each successive core update has gotten better at identifying and devaluing it.
The May 2026 update is the bill coming due for the “publish more, faster, cheaper” era.
Here’s the part that makes this cycle genuinely different from every prior core update.
For years, content marketers had an escape valve. If search traffic dropped, you could lean harder on social distribution. Different channel, different algorithm, different rules.
That escape valve is closing.
LinkedIn announced it’s targeting what it calls “AI slop,” low-effort, AI-generated content that sounds polished but offers little original thought. The platform’s detection system correctly flagged generic content 94% of the time in early tests.
Flagged posts don’t get removed. They get suppressed from recommendations, still visible to your direct connections but no longer spreading across the wider feed.
The targets are specific: engagement bait, recycled “thought leadership” with no original insight, and posts with obvious AI construction patterns.
And here’s the detail that should make every marketer who’s been leaning on AI templates wince. LinkedIn specifically called out the “it’s not X, it’s Y” format as an example of the formulaic AI content it plans to demote.
If you’ve read enough AI-generated LinkedIn posts, you know exactly the construction they mean. “It’s not about working harder. It’s about working smarter.” That cadence is now an active liability.
The crackdown extends to comments and video too. LinkedIn is building classifiers to detect bot comments based on language patterns and posting volume, and targeting “attention-bait videos” designed to hold attention without adding value.
The Replicability Test
The most useful frame to come out of this cycle is what practitioners are calling the “10-second test.”
If ChatGPT can reproduce your content in ten seconds, Google has no reason to rank it, and LinkedIn has no reason to distribute it.
That’s the whole thing, distilled.
For two years, the question was “can AI produce this faster?” The answer was almost always yes, which is exactly why the resulting content has no durable value. If a model can generate it on demand, there’s no scarcity, no differentiation, and no reason for any algorithm to privilege your version over the infinite identical versions.
The new competitive moat is content that does something rather than content that merely says something.
Content that says something: a generic explainer of a concept anyone can prompt ChatGPT to produce.
Content that does something: original research, proprietary data, an interactive tool, a genuinely contrarian analysis grounded in real experience, a case study only you could write because only you lived it.
The first category is now a liability. The second is the only thing that survives.
Why This Is a Multi-Front Problem
The reason this update matters more than a typical core update is the simultaneity.
In the past, AI content penalties were a search problem. You could treat them as an SEO issue, contain them to the organic channel, and route around them with paid, social, or email.
Now the same undifferentiated content is getting penalized in search and throttled in social at the same time.
If you built your SEO on scaled AI articles and your social presence on AI-generated LinkedIn posts, you’re not facing one problem with one fix. You’re facing the same root problem expressed across two channels, and the fix for both is identical: differentiation that AI can’t replicate.
That’s actually good news in a strange way. You don’t need two strategies. You need one, applied everywhere.
How to Audit Your AI Content Before the Next Update
If you’ve been scaling AI content, the worst move right now is to panic and start mass-deleting or mass-rewriting pages mid-assessment.
Google’s own guidance is to avoid reactive changes during and immediately after a rollout. Wait for the dust to settle, then assess with clear data.
Here’s a sane sequence.
Establish your baseline before you touch anything. Pull your Search Console data and annotate the May 21 to June 2 rollout window. You need to know exactly which URLs and which queries lost visibility before you start theorizing about why.
Apply the 10-second test to your top content. Take your highest-traffic pages and ask honestly: could someone reproduce this by prompting ChatGPT? If yes, that’s a page at risk, regardless of whether it dropped yet. The pages that survived this update are the early warning system for what the next one will reward.
Diagnose at the URL level, not the site level. Core updates rarely hit an entire site uniformly. Usually some content holds or gains while thin content collapses. You need to isolate which specific assets carry real differentiation and which are commoditized.
This is the kind of diagnostic Semrush One was built for. It lets you track ranking shifts at the keyword and URL level during the update window, benchmark your losses against competitors who may have gained, and see how your brand is being cited (or ignored) across AI search surfaces. The AI Visibility tools matter here because the same content quality signals that determine core update survival increasingly determine whether AI engines cite you at all.
They offer a free 7-day trial with access to 55+ tools if you want to run a full portfolio audit before committing.
Recommended reading: Semrush AI Visibility Toolkit Review: Does It Deliver?
Redirect production toward what AI can’t replicate. Once you know which assets are at risk, the path forward isn’t “stop using AI.” It’s “stop using AI as a strategy instead of a tool.” Use it to accelerate the production of differentiated content, original research, expert analysis, proprietary data, rather than to mass-produce the kind of content a model can generate on demand.
The Real Lesson
The uncomfortable truth is that none of this is a surprise.
Google has been signaling the death of scaled, undifferentiated content for two years. LinkedIn’s crackdown is the same logic applied to a different feed. The May 2026 update didn’t change the rules. It enforced rules that were already written.
The marketers getting hit this week aren’t being punished for using AI. They’re being punished for using AI to avoid the work that actually earns visibility, original thinking, genuine expertise, and content that exists for the reader rather than for the algorithm.
That work was always the moat. AI just made it briefly possible to fake your way around it.
That window is closed now. On both fronts.
Related Articles