Someone posts a hot take about startup culture. No capital letters. A missing space between two sentences. A dropped hyphen here and there. The writing looks casual, a little careless. But the argument underneath is unusually tight - four clean points, fully developed, with none of the wandering that real casual writing tends to have.
This is deliberate. People have figured out that polished, well-formatted writing now reads as AI-generated in many online contexts, so they're roughening their AI output before posting. The example making the rounds: startup advice that reads "chadgpt would tell you your saas idea has 14 direct competitors, your moat is nonexistent, and your timeline is delusional and then help you fix it anyway.that's not cruelty. that's what a good mentor does." Lowercase throughout, a missing space after the period, deliberate roughness layered over what is clearly structured, edited content.
Why the Surface Camouflage Works (Sort Of)
The original AI-detection heuristic was simple: AI writes in complete, correctly formatted sentences, uses balanced paragraphs, never makes typos, and tends to over-explain. Once enough readers internalized this, "polished = suspicious" became a real social signal in places like startup communities, academic forums, and creative writing spaces.
Introducing deliberate formatting errors disrupts that pattern. Lowercase is easy. Missing spaces and dropped commas take one second. For casual readers doing a quick scan, the roughness registers as human.
The problem is that formatting is a surface layer. The underlying content - the logic structure, the completeness of the argument, the vocabulary choices, the way points develop - is much harder to degrade without also degrading the output's usefulness. The deliberate-typo strategy produces writing that reads as casually careless but thinks unusually carefully, which is its own tell for anyone reading for more than a second.
The Transparency Problem
This behavior reflects a genuine tension. AI writing tools are widespread. Social norms about when it's acceptable to use them are still inconsistent, and in some communities still negative. So people improvise: use the tool, hide the evidence.
The transparency question is harder than it first appears. People don't disclose when they used spell-check or a calculator. Whether AI writing assistance crosses a threshold requiring disclosure depends heavily on context - and right now, different communities are reaching very different conclusions. The deliberate-typo workaround is what happens when people want both the output and the social acceptance of not having used it.