40% of Long LinkedIn Posts Are AI-Generated, Study Finds

What Happened?

Pangram Labs, a company specializing in detecting AI-generated content, published a report analyzing over a million posts on LinkedIn, X/Twitter, Substack, Medium, and Reddit. The main finding is alarming: 40% of long posts (over 250 words) on LinkedIn are entirely AI-generated. Across all platforms, one in four long posts was flagged as fully AI-written. LinkedIn not only leads in proportion but also concentrates 62% of all detected synthetic content, despite representing only a third of the posts analyzed. When including mixed content (AI and human), X/Twitter turns out to be the most saturated platform: nearly half of its articles (46.8%) are fully or partially AI-generated, with only 53.2% fully human. Substack, on the other hand, is the least affected, with a significantly lower proportion of synthetic content.

Why Is This Important?

LinkedIn promotes itself as a professional network where authenticity and credibility are key. That nearly half of its long-form posts are AI-generated undermines trust in the platform and raises questions about the quality of professional discourse. Additionally, the study found a paradox: users are more likely to use AI to speak in professional contexts tied to their real identity than on anonymous or informal platforms. This suggests that where reputation is most carefully managed, automation is most relied upon. This phenomenon is not new: in 2023, a Stanford University study already warned about the rise of AI-generated fake reviews on e-commerce platforms. The difference now is scale: with models like GPT-4 and Claude, synthetic content is nearly indistinguishable from human-written, making detection difficult and eroding trust in digital interactions.

Consequences for the Digital Ecosystem

The proliferation of AI-generated content has multiple effects:

• Content inflation: Signal-to-noise ratio decreases, making it harder for users to find genuine perspectives. On LinkedIn, this could lead professionals to ignore the platform due to lack of authenticity.
• Reputational risk: Brands and professionals using AI without transparency may be penalized by audiences or algorithms. A 2024 study by consulting firm Edelman showed that 76% of consumers distrust brands that use AI without disclosing it.
• Moderation challenges: Platforms will need to invest in detection tools and clear policies, something they have avoided so far. Reddit, for example, has already implemented mandatory labels for AI-generated content, while LinkedIn and X have not.
• Impact on SEO and visibility: Synthetic content can saturate search results, reducing the utility of platforms like Medium or Substack. Google updated its guidelines in 2024 to penalize AI-generated content without added value, but enforcement is inconsistent.

Additionally, the Pangram Labs report notes that long-form content is the most affected, suggesting users turn to AI to save time on posts requiring more elaboration. This contrasts with platforms like Reddit, where short posts and real-time discussions are less likely to be AI-generated.

What Should Readers Know?

The Pangram Labs report is a wake-up call for users and companies. It's not about demonizing AI, but demanding transparency. LinkedIn, X, and other networks still do not mandatorily label AI-generated content, leaving users without tools to discern. The solution is not to ban AI, but to encourage voluntary disclosure and develop authenticity standards. As Pangram Labs CEO Max Spero noted: “An internet completely flooded with undisclosed AI content is bleak, but we don't believe it's inevitable.”

For professionals, the recommendation is clear: use AI as an assistant, not a substitute, and always disclose its use. Companies, meanwhile, should establish clear policies on AI use in external communications. In the long run, trust in digital platforms will depend on their ability to balance AI efficiency with human authenticity.