sunrise, parabellum

Yuna, Grand Summoner

Kevin Glint: Instagram / Website / ArtStation

I want to talk about the harm caused by trying to police AI in creative communities, because things are getting really toxic. For the record, I dislike generative AI too. I think it’s built on theft and goes against the point of human expression. But turning online spaces into places where everyone is watched just to catch 'AI users' is hurting marginalized creators. Before you accuse someone based on a gut feeling or an unreliable detector, remember this: What people call AI 'tells' often look a lot like standard ESL writing. Non-native English speakers usually use structured, precise grammar or stick to common transitions because that’s how they learn the language. Accusing someone of using AI just because their writing sounds 'too clean' or 'formulaic' is often just xenophobia. This also hurts neurodivergent or hyper-focused writers. Many writers on the spectrum have their own structured styles, or they can write a lot in a short time. Calling high productivity or a different writing style 'robotic' pushes away the very people who make fan spaces interesting. The 'witch hunt' mindset does real damage. A community can handle a few cases of bad, AI-generated text getting through. What it can’t handle is a place where creators are too scared to post because a few loud voices are ready to attack them over a simple sentence.

AI models are trained on very structured, perfect grammar. So when a real person uses precise words, common transitions like 'Furthermore' or 'It is important to note,' or textbook grammar, detectors often say it’s AI. There’s a real ESL bias: studies show non-native English speakers get falsely accused much more often. Because ESL writers use standard words and sentence structures, detectors often flag their real work as robotic. One Stanford study found that 61% of essays written by international students were wrongly flagged as AI by at least one detector.