Who could have seen AI slop coming for LinkedIn? Anyone who has spent time there the past 20 years.

My colleague Sam Khoury lamented on LinkedIn about the AI slop on the platform after Fast Company reported on a study showing just how bad the problem is.

Fast Company wrote about a study from AI detection startup Pangram, and the article’s headline shares the claim that LinkedIn is “the most AI-saturated.” That’s misleading. Pangram only reviewed web-based, text-based platforms that could be tracked through its Chrome extension: LinkedIn, Reddit, X, Medium, and Substack. The report doesn’t get into all the slop on Instagram, TikTok, Snap, or even Facebook.

If you’re on LinkedIn, you’ve come across slop. The problem stems from LinkedIn’s incentives, and those incentives pre-date the dawn of GenAI.

How many connection requests do you get where they say, “We know 20 people in common, so we should connect”? It’s illogical. If that becomes the rule, then we will all become first-degree connections with each other, and the degrees will become meaningless. By knowing everyone, we know no one.

Most avid LinkedIn users have blown past Dunbar’s Number — the concept that we can have a working knowledge of 150 people at a time. Having some actual connection to someone matters, even if you met them once at a conference or they sat 50 desks away from you at a former company. Robin Dunbar himself elaborated that we can have different degrees of relationships, from best friends to people we recognize. We can’t recognize 8 billion people.

LinkedIn tends to reward short-term thinking. I often encounter obvious scammers who have a ton of mutual connections with me; many of my most esteemed friends seem to be lazy with vetting contacts. Since LinkedIn rewards reach, it's understandable that some would view every new connection as two new eyeballs to attract to their posts. 

LinkedIn’s algorithms are opaque, but there are some general rules that tend to work, and one of the most cited rules is to post regularly. That fuels even more perverse incentives to connect with scammers, but the reasons are more subtle.

If I accept an invite from Steve the Scammer (sorry to all the Steves reading this), I get one more automatic follower who might see my posts. But it's “better” than that. Steve the Scammer, who promises that he helps business owners like me get 5 to 10 new leads per week (what a guy!) and just helped a business like mine close $67K in sales last month (a genius!), is also likely to use AI (shocker!).

What's Steve going to do once we're connected? Well, nothing. Because his automation app will add me to his list of potential suckers, I mean clients.

Every so often, Steve's Scammerbot will automatically like my posts and contribute an AI-generated comment that will add oh-so-much value to the conversation. It's not slop, it's engagement! The more scammers’ connection requests I accept, the more these bots will flood my comments.

So what if my actual non-scammer contacts will ultimately wonder why all my friends are bots? My engagement score will go through the roof!

Could LinkedIn change those incentives?

Absolutely. In a heartbeat.

So much AI slop is so obvious that I’d wager 99 out of 100 of us would agree on many posts, and that’s without sophisticated detection algorithms. Plus, LinkedIn’s own tools offer AI assistance. It is fueling the problem directly all in the guise of helping its users come up with what to post and comment.

Is every media property that’s powered by user contributions the same? Not at all. Incentives vary everywhere, such as how outrage spreads fast on X, and BuzzFeed listicles still get love on Pinterest. The two systems aren’t interchangeable.  

Reddit hits different. Reputation scores aren't left up to a central algorithm; channel moderators can make or break someone's fate. You don't need 10,000 connections to create a post that a million people wind up seeing. You could follow half of Reddit's user base, and if your posts are boring, they're probably not going anywhere. Reddit has plenty of challenges, but its incentives for spreading AI slop are lower, and the disincentives are much higher compared to LinkedIn.

So what do we do with LinkedIn?

Not a lot. Unless you have some pull over the LinkedIn algorithm (Sam, have they given you the keys yet?). 

We’ll have to control what we can.

Connect with people you really know. Disconnect from spammers and scammers — and better yet, report them (it's so satisfying). Delete obvious AI slop comments you receive on your posts. Don't piss in the pool.

I deleted my X account proudly. I'm a nobody on Reddit, not so proudly. If we know each other, we probably have stayed in touch through LinkedIn at some point, and maybe often. LinkedIn is often the least bad option, and sometimes it's pretty good.

The one thing we can do is try to keep our part of the pool a little less warmer for everyone else.

— David Berkowitz, Chief Community Officer, Marketecture Media

Who: Media Buyers, Platform Strategy, Social Teams, Publishers

What: Speaking of incentives for Reddit, they’re questioning whether expanding AI data licensing could ultimately reduce traffic and advertising demand if users increasingly get answers from AI instead of visiting communities directly. The debate highlights how AI partnerships can create new revenue while threatening traditional advertising models.

Why it matters: Marketers relying on community platforms should pay close attention to how AI changes audience behavior rather than just media buying.

(Digiday)

Who: Agency Leaders, CMOs, Marketing Operations, CRM, Creative Strategy

What: Digitas CEO Amy Lanzi says AI’s biggest impact on advertising is making marketers faster and more effective. She argues agencies should use AI to streamline research, workflows, and execution while keeping strategy and creative judgment in human hands, and says marketing organizations are evolving toward more integrated, AI-enabled operating models.

Why it matters: Lanzi cuts against the prevailing AI hype. Instead of promising that AI will replace agencies or creative teams, she’s arguing that the competitive advantage comes from redesigning marketing operations around AI while keeping human judgment at the center.

(The Verge)

Who: Media Buyers, Programmatic Directors, Agency CTOs, Brand Marketers

What: The initial hype around agentic media buying has seemingly quieted down. While early tests like Butler/Till's autonomous campaign with PubMatic proved the technology works, the conversation has moved from flashy headlines to a dry, behind-the-scenes standards battle between Anthropic’s open Advertising Context Protocol (AdCP) framework and the IAB Tech Lab's competing framework (AAMP). Despite the lack of public chatter, agencies like Omnicom continue quietly running live agent-to-agent buys for major clients.

Why it matters: Agentic advertising is transitioning from a hyped-up revolution to a quiet, incremental reality; marketers should expect tools to continuously streamline media planning behind the scenes rather than replacing planners overnight.

(MediaPost)

Liking AI Brief? The Refresh is a weekly recap of all the most important advertising news, written by AdTechGod.

Thanks for reading this week’s AI BRIEF by Marketecture. 

Marketecture Media is committed to bringing you the latest news, opinions, and community updates. 

If you are interested in advertising on this newsletter, our podcasts, The Advertising Forum, or in our Slack community, you can contact us here.