In the wild · the comment sections
AI in social media comments
30-second gist~30s read
Bot accounts have always existed on social media. They used to be obvious — broken English, copy-pasted text, weird usernames. Now an AI can write comments that pass casual reading: contextually relevant, well-spelled, on-topic.
They're used to push narratives, harass people, sell products, and inflate engagement metrics. The point of seeing them isn't paranoia — it's recalibration. Comments are a noisier signal than they used to be.
If you want more
How to spot one
- Short comment history. Account opened in the last few weeks, mostly active in one thread.
- Replies that fit the topic but don't really engage. Generic agreement or generic disagreement, no specific reasoning.
- Sudden link-out to a product, an article, a campaign.
- Profile picture too perfect — often AI-generated.
- Cluster behaviour. Several accounts replying in the same shape on the same thread within minutes.
What this means for online discussion
The fix isn't to suspect every comment. It's to weight comments less heavily as a signal. "Lots of people are angry about this" used to mean something. Now it sometimes means "a campaign paid for fifty replies". Real signal still comes from people you already trust, journalists with names, and conversations that go beyond a single thread.