The Gen-AI Police

“Stop using AI, Greg!”

It’s 2026. You’re on the train Monday morning, commuting to the office. As usual, you open LinkedIn and look for something useful to read before work. But before you even process any great ideas from Greg (well, your Android TPM @ iAI.com), your brain starts analyzing his writing…

The goddam em dash.
The “not A, but B” structure.
The clean little bullet list. (just like this one, ha!)
The AI tone that sounds polished but oddly empty.

At this point, the reading experience changes. Instead of trying to understand people’s points, you are constantly asking, “Was this AI-generated?” And if the answer feels like yes, you scroll away. “Yeah. AI shit.”

It gets worse when it is someone you know. You see a post from a coworker, a former teammate, or someone in your whatever networks, and the whole thing sounds like it came out of the same LLM. You start typing: “Greg, this is 100% AI…” But you are too nice to actually hit that send button to leave it in the comments, so you just scroll away.

This is a small moment of annoyance, but something bigger is happening. Generative AI has added a new layer of verification to everyday content consumption. I call us the AI police.

The investigation

The same thing is happening with images and videos. A image or video post on X is no longer just a post. People look at the hands, the shadows, the background, the anatomy, the source, the comments, and then sometimes ask Grok whether the thing they are looking at is real (ironically).

I’ve seen this happen in smaller communities too. On Fishbrain, people post catch photos. This whole thing should be simple: someone caught a fish, took a picture, and shared it with other people who like fishing. But now, fake AI catch photos show up, and the comments turn into an investigation. “I was at Lake Halcyon at 2 PM. Did not see you.” “Did you actually catch this fish?”… A hobby feed becomes a police station.

This is the interesting part. It’s fine that AI content exists. The problem is that the audience now has to investigate the content before trusting it. The user cannot simply consume anymore.

The cognitive load

A lot of platforms are optimized for content volume. More posts, more images, more videos, more engagement, more things for endless scrolling. But from the user perspective, it is not simply “more content.” They’re experiencing more uncertainty, looking for trust signals much more than before.

More content does not automatically mean more value. In lots of cases, it means more work from users. The reader has to decide whether an article is worth reading. The viewer has to decide whether an image is real. The community has to decide whether a post is verified. And the audience becomes the first moderation layer.

That creates the new cognitive load. Before generative AI, the mental model was simple: “Do I care about this?” But now it can look more like:

“Do I care about this?”
“Is this real?”
“Was this written by a real human?”
“Is this image generated?”
“Is this post farming clicks?”
“Is this worth trusting?”

That is a lot of work before the user even gets to the actual content. And once this behavior becomes normal, it affects everything around us, not just the AI-generated content. A human-written post can start to feel suspicious just because it uses the same structure as AI writing. A real photo can get questioned because people have seen so many fake ones. This is truly killing me. I have even seen people spend time making sure they don’t sound like AI even they wrote the piece themselves.

Authenticity becomes part of the interface

This is where the product problem starts. Authenticity will be playing an even larger part of the interface. More important than ever before. A feed, a profile, or even a piece of ad, are more than what they were. They carry credibility signals. They can change whether someone stays or leaves.

And this burden should not sit entirely on the user. If platforms allow AI-generated or AI-assisted content, they need to think carefully about how trust is communicated. The goal is legibility. Users should not have to become detectives every time they open a feed. Platforms can reduce that verification labor with clearer signals: where the content came from, whether AI was involved, who posted it, how reliable the account has been, and what the community has already found. Not all of this needs to become a huge roadmap initiative. A small signal can be enough. Maybe a community norm is enough. Sometimes maybe the right answer is stricter moderation.

Regardless, the design responsibility is real. Trust has become harder to maintain in the era of gen-AI. For designers, product builders, and content creators.. How do we build the right thing and make sure people still trust what they are looking at?