AI-generated posts flood social media, LinkedIn hit hardest

LinkedIn has become the clearest test case for trust in the age of synthetic content. Pangram, an AI-detection company, said the professional network accounted for nearly two-thirds of the AI content it detected across social platforms, even as its report found the average AI rate across all scanned items was 13.8%.

Pangram published the report on July 9 after analyzing roughly 1 million posts from LinkedIn, Medium, X/Twitter, Reddit and Substack over a two-month period. The company said the data came from its Chrome extension, which lets users scan posts as they scroll and only contributes statistics if they opt in anonymously. Its findings showed one in four longform social posts were fully AI-generated, and more than 40% of longform LinkedIn posts were flagged that way. In the same dataset, nearly half of X/Twitter articles contained AI writing, while even Substack, the least AI-heavy longform platform in the sample, still had more than a fifth of posts flagged as AI-generated or AI-assisted.

The pattern pointed to a deeper workplace problem than spam alone. LinkedIn is built on professional credibility, personal branding and proof of expertise, which makes AI-written filler harder to spot and easier to mistake for genuine insight. Pangram said longer posts were more likely to be AI-generated than shorter ones on four of the five platforms it studied, a warning that the most polished-looking content may also be the most manufactured.

Max Spero, Pangram’s chief executive and co-founder, said the company’s browsing-based method mattered because it captured what people actually see while scrolling rather than the broader internet. He called AI content “a tax on readers’ time.” That scrutiny has widened as LinkedIn expands its use of user data for AI features and training, a practice that has drawn privacy criticism and raised fresh questions about how much of a user’s professional life is feeding the same systems now shaping the feed. LinkedIn has also moved to suppress generic AI-generated posts in recommendations rather than remove them outright, an acknowledgment that the flood of machine-made material is becoming a feed-quality problem as much as a technology problem.