Welcome to a new week of updates. LinkedIn says it’s going to start reducing the reach of AI-generated posts, comments and engagement bait in users’ feeds.
LinkedIn’s Laura Lorenzetti said the company wants to prioritize “real voices, authentic perspectives and lived expertise” over generic AI-written content.
The platform says it has built systems to detect AI slop, including repetitive thought leadership posts, automated comments and responses that restate someone else’s post without adding anything new.
Content flagged as overly generic will be shown to fewer people outside a user’s immediate network. LinkedIn says early testing correctly identified generic content 94% of the time.
Let’s get into what else is new.
LinkedIn launched a new comparison system called Crosscheck that tests how AI models perform real tasks.
Crosscheck measures things like writing quality, factual accuracy, reasoning and usefulness between different LLMs. Users submit prompts and see responses from two different models without knowing which company made them. They then choose which answer they prefer.
LinkedIn says this creates more realistic feedback because people judge outputs based on usefulness rather than brand reputation. The company plans to use the data to build leaderboards showing which models perform best for different kinds of work and industries.
The platform also expanded its partnership with ad verification company DoubleVerify so brands can measure insights across its Audience Network like fake traffic, brand safety and geographic accuracy after ads run.
This should give brands more insights and allow greater transparency around what they’re paying for, the company said.
X
X says unverified accounts are now limited to 50 original posts and 200 replies per day, while all accounts can send up to 500 DMs daily. They said some of the limits are also broken into certain time windows throughout the day to prevent spamming.
Accounts following more than 5,000 people may face additional restrictions based on their follower-to-following ratio. X says accounts which aggressively follow users, mass-message identical DMs or post repetitive content may temporarily lose certain capabilities.
They have also added a new metric that shows how many of an account’s followers are actually active on the platform. Nikita Bier, X’s head of product, said the display should provide more context around audience size and performance by separating total follower counts from followers who regularly use the app.
Reddit is adding some new ad tools to help app marketers drive downloads and sales. This includes a new dashboard that lets advertisers compare Reddit’s ad performance data with outside tracking tools like AppsFlyer and Apple’s SKAdNetwork.
The company says app install volume from Reddit ads grew 129% over the last year, and that Reddit users are higher-intent consumers who spend more time in apps.
Snapchat
Similarly, Snapchat added Unified Attribution, a tool that combines Snapchat ad data with third-party measurement tools into one dashboard.
Advertisers will be better able to understand what’s driving downloads, purchases and other business results across campaigns, the company said.
Unified Attribution will combine the insights into one view so marketers can make more informed decisions about where to spend money and how campaigns are performing.
Threads
Threads is expanding its brand safety controls so advertisers have more say over where their ads appear.
Brands can now use third-party block lists to avoid having ads placed next to content they consider risky or inappropriate. This feature is already available on Facebook and Instagram, and now includes Threads.
Courtney Blackann is a communications reporter. Connect with her on LinkedIn or email her at [email protected].
