AI-Generated Posts Overrun Social Media: A Serious Problem on LinkedIn and X

A survey by the AI detection platform Pangram has revealed that one-quarter of long-form posts on social media are entirely generated by AI. The proliferation of AI-generated content is particularly pronounced on Microsoft-owned LinkedIn and Elon Musk’s X (formerly Twitter). According to a report by Brandon Vigliarulo of The Register, Pangram analyzed over one million posts via a Chrome extension released at the end of April 2026.

Pangram offers a Chrome extension for a $20 monthly subscription that automatically scans users’ feeds on LinkedIn, Medium, Substack, X, and Reddit, detecting AI-generated or AI-assisted content. Since its launch in April 2026, it has analyzed over one million posts from users who opted in to share data, revealing that AI-generated content has penetrated deeply, especially in long-form posts.

Pangram defines a “long-form” post as one exceeding 250 words. According to its analysis, 25% of long-form posts across all surveyed platforms were entirely AI-generated. This figure does not include posts created with AI assistance (so-called partially AI-generated content). In other words, even when excluding cases where users used LLMs (large language models) to embellish their writing, this percentage remains.

Status by Platform

Disparities between platforms were notable. LinkedIn was the most severe, with 41% of long-form posts judged to be AI-generated. This is a decline from the 54% reported by the AI detection company Originality.ai in late 2024, but still a high level. However, Originality.ai defined long-form as 100 words or more, and differences in criteria may affect the numbers.

On LinkedIn, even short-form posts (50–250 words) were 30% fully AI-generated, highlighting that many long-form posts posing as thought leadership are heavily reliant on AI. Interestingly, only 4.3% of long-form posts on LinkedIn were AI-assisted (partial use). Users tend to either entrust everything to AI or not use it at all, showing a polarization. Only 55.2% of long-form posts were actually written by humans.

The situation on X is similar. 25% of long-form posts are fully AI-generated, and an additional 23.2% are estimated to be AI-assisted. Human-written posts accounted for only 52.7%. On X, the proportion of partial AI use is higher compared to LinkedIn, suggesting that users may lean toward using AI for proofreading or idea generation.

Combining both platforms, roughly half of all long-form posts involve AI in some form. Users scrolling through their feeds are confronted with content that may or may not be written by humans.

Survey Methodology and Reliability

Pangram’s Chrome extension operates in an environment where users voluntarily agree to share data. The sample size is large—over one million posts—but it is important to note the bias toward opt-in users. Users with a higher interest in AI-generated content may be more likely to install the extension, meaning the actual AI generation rate could be lower than the survey results.

Nevertheless, the trends align with similar surveys by Originality.ai in 2024, confirming that the increase in AI-generated content is a definite trajectory. LinkedIn, in particular, is a platform where professionals showcase their expertise, posing a significant risk of spreading “fake insights” embellished by AI.

Challenges Facing the Industry

The flood of AI-generated content poses a serious problem for platform operators. If human-to-human trust-based interactions are replaced by automatically generated, hollow text, the value of the platform itself declines. Microsoft dominates the business SNS market through LinkedIn, but a drop in quality could lead to user attrition, a risk that cannot be ignored.

For X, Musk’s acquisition brought major changes to the platform’s content moderation policies. The increase in AI-generated content could be seen as one result. However, Pangram’s survey only judges the method of generation, not the quality of the content itself; it is important to note that AI-written posts are not necessarily all low-quality.

On the other hand, the accuracy of AI detection technology itself is open to debate. Independent verification is needed to determine how accurate Pangram’s judgments are, and the possibility of false positives should be considered. As AI models rapidly improve—such as OpenAI’s release of the GPT-5.6 series—the cat-and-mouse game between detection and generation will continue.

Editorial Opinion

In the short term, detection technology and countermeasures for AI-generated content will likely become factors that determine competitive advantage among platforms. LinkedIn, in particular, relies on its credibility as a business-oriented SNS, and may face pressure to implement measures such as visualizing AI posts or mandatory labeling. Similarly, X will need to take action to maintain advertiser trust. How each platform handles AI-generated posts over the next three to six months will be closely watched.

From a long-term perspective, we have entered an era where the very reliability of information on social media is being questioned. If AI-generated content becomes indistinguishable from human posts, the value of platforms may shift from “who wrote it” to “what is true.” Over a one- to three-year span, the need for regulations mandating disclosure of AI-generated content and mutual certification systems between platforms is likely to grow. The editorial team also considers improving user literacy to be important. Rather than relying solely on AI detection tools, users should cultivate the habit of verifying sources and logical consistency on their own.

References

• “AI slop writing has taken over the internet, particularly LinkedIn and X”, by Brandon Vigliarolo — The Register, 2026-07-09T21:01:04.000Z (ARR)
• Source URL: https://www.theregister.com/ai-and-ml/2026/07/09/ai-slop-writing-has-taken-over-the-internet-particularly-linkedin-and-x/5269525