LinkedIn is taking another step toward cleaning up its feed—this time by giving users a more direct way to flag what it’s calling “AI slop,” and by changing the way its own writing assistance works. The move is notable not just because it adds a new reporting button, but because it signals a broader shift in how major platforms are trying to manage generative AI content: less about trying to perfectly detect every AI-generated post, and more about building feedback loops that help the system learn what users consider low-value.
At the center of the update is a new reporting option that appears alongside other moderation tools. LinkedIn will let people report posts with a label along the lines of “seems like AI slop.” The phrasing matters. It’s not “AI-generated” in general, and it’s not “misinformation” or “spam” in the traditional sense. Instead, it targets a specific user experience problem: content that looks machine-produced, feels generic, and adds little signal—posts that may be technically compliant but still degrade the quality of the conversation.
This is a subtle but important distinction. Many platforms have historically relied on categories like spam, harassment, or policy violations. Those are necessary, but they don’t always capture the everyday annoyance users feel when they scroll past repetitive, templated, or overly polished text that seems designed to game engagement rather than contribute insight. “Slop” is a term that has emerged across the internet to describe exactly that kind of low-effort output. By turning that concept into a reportable action, LinkedIn is effectively acknowledging that “quality” is not only a matter of legality—it’s also a matter of usefulness.
The second part of the change is equally telling: LinkedIn is replacing its own AI writing feature with a proofreading tool. In other words, instead of offering a capability that helps users generate text from scratch, LinkedIn is moving toward assistance that refines what a user already wrote. Proofreading tools typically focus on grammar, clarity, tone, and structure—helping the author improve their draft rather than producing a full post that may sound plausible but disconnected from the author’s real perspective.
That shift aligns with a growing consensus among platform operators: generation at scale can flood feeds with content that is fluent but shallow. Proofreading, by contrast, tends to preserve authorship. It encourages users to start with their own ideas and then polish them, which can reduce the volume of posts that feel interchangeable. It also changes the incentives. When a tool can generate an entire post, it becomes easier for someone to produce many variations quickly. When the tool is limited to editing, the bottleneck becomes the user’s own thinking and drafting.
Taken together, these updates suggest LinkedIn is trying to address two sides of the same problem. The reporting button gives LinkedIn a way to collect labeled examples of “low-quality AI behavior” from real users. The proofreading tool reduces the likelihood that LinkedIn’s own product will contribute to the creation of mass-produced content. It’s a combination of demand-side moderation (what users flag) and supply-side design (what the platform encourages people to do).
Why a “report” button is a big deal—even if it sounds small
On the surface, adding a button might look like a routine trust-and-safety tweak. But in practice, reporting tools are one of the most valuable inputs a platform can get. They convert subjective impressions into structured data. And when the label is specific—like “seems like AI slop”—it can help train systems to recognize patterns that automated detection alone might miss.
Automated moderation can struggle with nuance. A post can be AI-assisted and still be thoughtful. A post can be human-written and still be generic. A post can be short and still add value. So if LinkedIn tries to build a binary classifier for “AI-generated vs not,” it risks false positives and false negatives. Users may report posts that are merely poorly written, or systems may fail to catch posts that are AI-generated but not obviously so.
By contrast, a “slop” report is closer to the user’s actual complaint: the post doesn’t feel like it belongs in a professional network. It’s not necessarily about whether the words were produced by a model; it’s about whether the post meets the standard of contribution that LinkedIn users expect.
That means the label can become a proxy for multiple underlying issues: repetitiveness, lack of specificity, generic motivational language, vague claims without evidence, and the “same template, different industry” effect that many users have started to notice. Over time, those reports can help LinkedIn refine ranking and moderation policies, not just remove content but also adjust how content is surfaced.
There’s also a strategic advantage. Reporting is a form of lightweight community governance. It distributes the work of identifying problematic content across millions of users, rather than relying solely on internal teams or automated systems. Of course, that introduces its own risks—brigading, misuse, or inconsistent judgments—but platforms typically mitigate this by aggregating signals, weighting reports, and using additional context.
Still, the key point is that LinkedIn is explicitly inviting users to participate in defining “quality.” That’s a powerful move, because it acknowledges that the platform’s definition of “good” content is ultimately shaped by user behavior and expectations.
The proofreading pivot: less generation, more refinement
The replacement of LinkedIn’s AI writing feature with a proofreading tool is not just a product change; it’s a philosophical one. Generation tools can be useful, but they also make it easier to produce large volumes of content quickly. In a feed-driven environment, speed and volume can translate into visibility. Even if the content is technically allowed, the net effect can be a feed that feels crowded with low-signal posts.
Proofreading tools, on the other hand, are typically constrained to improving text that already exists. That constraint matters. It reduces the ability to mass-produce posts that sound “right” but aren’t grounded in the author’s experience. It also encourages a workflow where the user writes first, then uses AI to polish.
From a user perspective, this can feel like a downgrade if someone wanted a full “write my post” assistant. But it can also feel like a better fit for LinkedIn’s culture. LinkedIn is built around professional identity—roles, industries, achievements, lessons learned. A proofreading tool supports that identity by helping users express their own thoughts more clearly, rather than outsourcing the entire voice.
There’s another angle: proofreading tools can be tuned to reduce common AI artifacts. For example, they can encourage more concrete language, remove overly generic phrasing, and suggest structure that reads like a human authored it. While proofreading isn’t a guarantee against low-quality content, it can steer users away from the most obvious “model-like” patterns.
If LinkedIn is serious about reducing “slop,” this is a sensible direction. It’s harder to generate a convincing, specific narrative without having something real to say. Proofreading doesn’t create that narrative; it helps the user present it.
A unique take: LinkedIn is shifting from “AI as a writer” to “AI as a quality editor”
Many platforms have treated AI writing as a productivity feature: type a prompt, get a post. That approach is attractive because it’s immediate. But it also turns writing into a commodity. If anyone can generate a post in seconds, the differentiator becomes who can generate faster—or who can generate more.
LinkedIn’s move suggests it wants to keep the differentiator as the author’s perspective. The platform is essentially saying: we’ll help you edit, but we won’t replace your voice. That’s a subtle but meaningful repositioning.
It also changes how users interact with AI. Instead of asking AI to invent, users are nudged to draft. That can lead to better outcomes even without any sophisticated detection. When people write first, they naturally include details they care about: their own metrics, their own lessons, their own context. Even if they use AI to clean up the language, the content is more likely to be anchored in reality.
This is where the “slop” report and the proofreading tool connect. The report gives LinkedIn a way to identify posts that feel unanchored and generic. The proofreading tool reduces the likelihood that LinkedIn’s own product will produce unanchored content at scale.
What could go wrong: ambiguity and the risk of over-reporting
Any system that relies on user judgment faces a challenge: people disagree about what counts as “slop.” Some users may interpret “slop” as “anything I don’t like.” Others may report posts that are simply poorly written, or posts from creators whose style differs from their own. There’s also the possibility of strategic reporting—competitors or bad actors flagging content to reduce visibility.
LinkedIn will likely need to handle this carefully by treating reports as signals rather than instant verdicts. Typically, platforms aggregate reports, look for patterns across many users, and cross-check with other indicators such as engagement anomalies, repetition across accounts, and similarity to known low-quality templates. The goal would be to avoid punishing legitimate creators who happen to use AI tools responsibly.
Another risk is that “slop” can become a catch-all category. If the label is too broad, it may capture content that is not actually low-quality but is perceived as such due to topic sensitivity, political disagreement, or niche writing styles. LinkedIn will need to ensure the reporting flow includes enough context or guidance so that users understand what they’re flagging.
Even if LinkedIn does everything right, there’s a cultural risk: users may start policing each other’s writing. Professional networks can become echo chambers, and moderation labels can amplify that dynamic. The best-case scenario is that the “slop” report becomes a tool for removing genuinely low-signal content, not a weapon for taste-based disputes.
How this could affect feed ranking, not just takedowns
One of the most interesting parts of this update is what it implies about LinkedIn’s ranking strategy. Reporting buttons are often thought of as moderation