LinkedIn is rolling out a 'Seems like AI slop' report button, letting users flag posts that read as machine-written after AI detector Pangram found 41% of longform LinkedIn posts came back as completely AI-generated. Chief Product Officer Hari Srinivasan announced the change on Thursday alongside a broader package of measures aimed at cutting synthetic content from the feed. The button now appears in the three-dot menu on posts and hides flagged content once submitted.
The Microsoft-owned network is also training new classifiers to detect AI slop and other low-quality content, which will suppress how often those posts surface in suggested content from outside a user's network. LinkedIn says it now blocks hundreds of thousands of automated comment attempts every day, and millions of automation attempts over the past couple of months.
“AI slop is a top priority for all of us. We really care about this.”— Hari Srinivasan, LinkedIn Chief Product Officer
Srinivasan framed the slop problem as a threat to what LinkedIn sells advertisers and recruiters: a network of real professionals writing in their own voices. The company is going a step further and killing its own 'enhance your post' feature, which used AI to rewrite drafts. It will be replaced with a tool that proofreads text without changing the author's voice.
Key facts
- 01LinkedIn added a 'Seems like AI slop' report button on Thursday, July 30, 2026, letting users flag posts that read as machine-written.
- 02AI detector Pangram found 41% of longform LinkedIn posts were flagged as completely AI-generated.
- 03LinkedIn now blocks hundreds of thousands of automated comment attempts daily and millions of automation attempts in the past couple of months.
- 04The Microsoft-owned network is killing its 'enhance your post' AI writer, replacing it with a proofreader that preserves the user's voice.
- 05Pangram announced $9M in new funding this week to expand its AI-content detection work.
LinkedIn is also planning to privately flag posters in their own dashboards when their content is coming across as inauthentic due to heavy AI use. The idea is to nudge borderline users — the ones polishing genuine writing with AI — before their posts start getting reported by others. Full-on slop, by contrast, gets pushed down the algorithm.
The report button is not just a moderation lever. Srinivasan said it will help LinkedIn 'tune our models and make better feeds' by giving the classifiers a live signal on what users consider synthetic. Every flag is training data.
The move lands in a week where AI-slop detection is suddenly a business. Pangram, the detector that produced the 41% figure via 404Media, announced $9 million in new funding this week. Substack added a Pangram-powered tool last week to help readers see when a post on its site was AI-written.
“People come to LinkedIn to connect with real people and share their real perspectives, ideas, and expertise.”— Hari Srinivasan, LinkedIn Chief Product Officer
Other platforms are visibly struggling. Digg had to shut down its Reddit competitor in March, saying it could not get ahead of the bot traffic flooding the site. Cloudflare has said automated bot traffic now exceeds human requests on the web, a threshold it reached earlier than its own forecasts predicted.
LinkedIn is a specific target for slop because the incentives point that way. Long, earnest, first-person posts drive reach on the platform, which makes them the exact shape of output that a generic large language model produces on request. The result is a feed increasingly full of confident, mid-length essays about leadership, resilience, and 'lessons I learned' — many of which now trip AI detectors.
The obvious limit here is detection accuracy. Pangram's 41% figure is a detector's classification, not a confession, and even well-tuned classifiers produce false positives on human writing that happens to be formal, templated, or edited. LinkedIn has not published its own precision and recall numbers for the new classifiers, and the report button will produce noisy signal from users who dislike a post for reasons unrelated to AI. Overreach could punish legitimate posters who use AI as an editor rather than a ghostwriter — which is why the private dashboard flag exists as a softer intervention.
For LinkedIn, the calculation is straightforward: the network's monetizable value comes from people trusting that the person on the other end is a person. If professional social feels like a bot farm talking to itself, ad rates and recruiter subscriptions follow the trust curve down. Expect every consumer platform whose product is 'real humans talking' — Substack, Reddit, and X among them — to end up shipping some version of this button within the next year, and expect the detector-vendor market that Pangram is building into to consolidate quickly around whichever tool the biggest platforms end up licensing.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
