AI is Killing Fan Translation. The Fansubbers Couldn’t Be Happier.

You know the panic. Every time a new AI model drops, a fresh wave of creators trembles. Artists fear their styles will be stolen. Writers fear their words will be automated. The narrative is always the same: the machines are coming for our passions, and there’s nothing we can do about it.

But I recently stumbled into a corner of the internet that has a radically different reaction to the AI revolution. They aren’t just unafraid—they are actively cheering for their own obsolescence.

They are the fan translators. The subbers and scanlators who spend hundreds of unpaid hours translating Japanese anime, variety shows, and manga for fans who can’t speak the language.

I assumed they would be bitter. I assumed they would view AI translation tools as the ultimate betrayal. Instead, I found a community quietly preparing to write its own graduation letter.

Take Eliza. She runs a fan subbing group for a popular Japanese actor. When the actor went on a two-year hiatus, the old subbing groups disbanded. When he finally returned to TV, there was no one left to translate. Eliza doesn’t speak Japanese—she describes her proficiency as ‘Duolingo level.’ But she wanted to watch the show. So, she started recruiting.

She eventually scraped together a team of 25 people. But here’s the dirty secret of fan translation: it’s mostly grueling, mind-numbing labor. Before you even translate a word, you have to ‘time’ the subtitles. Someone has to manually mark the exact millisecond a sentence starts and ends to match the audio. For a 20-minute video, this takes three to five hours.

Then, a programmer joined her group and brought AI with him. He built a custom app that automated the transcription, timing, and initial translation. The three-to-five-hour grind? The AI does it in 20 minutes.

I asked Eliza what happened to the people who used to do that manual labor. She said they mostly just hang out in the Discord now. Over half the group does absolutely no work. And she doesn’t care.

The community was never the output. The output was just the excuse to build a community.

Eliza’s group chats late into the night about the show, life, and everything in between. They pay out of their own pockets for server costs and AI subscriptions just for the privilege of working for free. The tedious labor was never the point. The shared love was the point.

Then there’s Zhenyu, a programmer who calls himself a ‘single-person AI roasting machine.’ (In fan circles, untranslated videos are ‘raw meat,’ and adding subtitles is ‘roasting’ them.)

Zhenyu loves a relatively niche streamer who has no dedicated translation group. Before AI, he would just wait and hope someone else would translate it. Usually, no one did. So, he built his own AI pipeline. After a two-hour livestream ends, he feeds it to his tools, does a quick manual check on names and timing, and uploads the subtitled version before sunrise.

He doesn’t speak Japanese either. When I asked if he worries about the AI making translation errors, his response was brilliant: ‘If someone complains about my translation, it means the streamer finally has a subbing group.’ He isn’t protecting his territory. He’s just trying to fill a void until the void can fill itself.

But what about manga, where the art of ‘typesetting’—replacing Japanese text with English while matching the font, style, and background art—requires a deeply human touch?

Anbian runs a manga scanlation group. He tested every AI tool on the market. They were fast, but they were terrible. The AI didn’t understand how to redraw the art hidden behind the text bubbles. It didn’t know how to choose the right font to convey a character shouting versus whispering. One of his typesetters has 575 different fonts installed on her computer to match the exact vibe of a scene.

But Anbian isn’t anti-AI. He knows the tools will eventually get good enough to handle the art, too. And when that day comes, he’s ready to step aside. He compares his group to a college club. People join to learn, to practice Japanese, to hang out. But eventually, everyone graduates.

We thought AI was coming for the creators. It turns out, it’s just taking out the trash—the tedious, soul-crushing chores that kept them from actually enjoying what they love.

When I asked Eliza what she would do if, five years from now, you could just click a button on any video and get a perfect, emotionally resonant translation instantly, she didn’t hesitate.

‘That would be great,’ she said. ‘Everyone can finally eat well.’

That’s the twist the tech doomers don’t understand. The doomsayers look at fan translation and see a ‘workflow’ being disrupted. They see efficiency replacing effort. But they miss the entire point of why these communities exist.

If the community dies, the first ones to celebrate will be the ones who built it. Because their ultimate success was making themselves unnecessary. They didn’t want to be the gatekeepers. They just wanted everyone to be able to watch the show.

If your community’s only value is the tedious labor it performs, AI will destroy it. But if its value is shared identity and love, AI just hands you back your time.

The fansubbers aren’t dying. They’re just finally graduating.