Platforms Built For Human Connection Filling Up With Content That Isn't

Research from Pangram, a technology company that built a browser extension for detecting AI-generated writing, found that 41% of long-form posts on LinkedIn show indicators of AI generation, the highest rate of any major social platform tracked. X follows with around a third of long-form posts showing similar signals. Pangram's data is based on average measurements across platforms, and the company acknowledges no AI detection tool is 100% accurate.

Every major platform is dealing with the problem to varying degrees. Pinterest is the only one to have introduced an opt-out mechanism, giving users the ability to limit AI-generated content in their feed. Reddit recently shared details of its efforts to combat a rise in AI-generated comments. X head of product Nikita Bier announced changes to creator monetization incentives in March specifically to address a spike in AI-generated deepfakes tied to coverage of the war in Iran. LinkedIn has responded with reach penalties when AI spam is detected.

The challenge is compounded by the fact that the same platforms investing in AI suppression are also heavily invested in AI development. Meta has spent hundreds of billions on AI infrastructure and is actively prompting Facebook and Instagram users to generate images and compose posts using its tools.

When Instagram announced the integration of Meta's Muse image generation tool, user replies were described as almost universally negative, with responses including requests for an option to turn AI-generated content off entirely.

Full story: Social Media Today