Source:
The RegisterOr: Shock as users of world's largest slop repository report lots of slop More than 1 million LinkedIn users have clicked the newish “Seems like AI slop” button since the social network launched it at the end of July. Hari Srinivasan, LinkedIn’s chief product officer, said that compared to the period before launching the AI slop reporting button, those who copy-paste AI-written posts are seeing around 40 percent fewer views. “Despite the progress, we know we have more to do to ensure LinkedIn remains a place where you can find real people and real perspectives... this all remains very top of mind,” he said. No single piece of feedback is used to adjust how widely a post travels, inside or outside a user’s network, he said, and distribution is determined by a range of signals. Srinivasan added that users will also start seeing messages in their analytics dashboards to understand how viewers are receiving their posts. If people think there’s too much AI use, the platform will relay that to the user. Any logged-in user can report a post for suspected AI slop using the platform’s three-dot option menu. The CPO made it clear in his initial announcement of the reporting button that there is a difference between AI and “slop.” LinkedIn believes people have a right to continue using AI to, ironically, make themselves sound more authentic. Regardless of what some may think about the button being used to train AI itself - rather than LinkedIn’s users - on how to sound more human, the company also wants the white-collar social media tryhards making these posts to get human feedback instead of an automated AI detector. Srinivasan said that, behind the scenes, LinkedIn is working on an array of automatic detection mechanisms for AI slop content. Prior to launching the slop button, the platform had used these detections to catch hundreds of thousands of AI-slop comments each day, and “billions of other automation attempts,” such as slop posts and posting at scale, “in the last couple of months alone.” As part of the war against slop, Srinivasan also announced in July that LinkedIn was removing its AI-driven “enhance your post” feature that made posting slop all the more simple. It replaced the feature with a proofreader. In announcing “slop” as its word of the year for 2025, Merriam-Webster noted that the term relates to a broad range of lazily compiled content. Fake news can be slop. Social media posts, on LinkedIn and X especially, can be slop. Propaganda spread by the US president can be slop. Institutions as treasured as UK pubs stand accused of sloppy practices, and pretty much everything in between – AI slop pervades the modern world. In the security space, AI slop bug reports have been a recurring issue. The nature of the headache, however, has changed. As Daniel Stenberg, founder and lead developer of cURL, explained earlier this year, AI-generated vulnerability reports have become a serious drain on maintainers' time as they often look credible at first glance but turn out to be baseless once investigated. Now, though, maintainers are struggling to tackle the high volume of high-quality bug reports that are made with the help of AI – no slop. ®
Get AI news in your inbox
Daily digest of what matters in AI.