Does Turnitin's AI detector work on text that was translated from another language?

[Eric]

This is kind of a weird question but hear me out.

I have a friend (actual friend, not me) who's an international student. English is his third language. He writes his papers in his first language (Spanish) and then uses DeepL or Google Translate to convert to English, then does some editing.

His last paper got flagged by Turnitin's AI detector as "45% AI-generated." He swears he didn't use ChatGPT or anything—just translation tools.

Here's my theory:

Translation tools use AI. Like, DeepL is literally neural machine translation. So when he runs his Spanish through it, the output is technically "AI-generated" even though the IDEAS are his. But Turnitin can't tell the difference between "AI wrote this from scratch" and "AI translated this from human writing."

What we need to know:

• Does Turnitin flag translated text as AI?
• If so, how do international students prove it's translation not generation?
• Is there a "safe" way to translate without getting flagged? (Like, translate in smaller chunks and heavily edit?)
• Has anyone successfully appealed an AI violation by showing original language drafts??

The stakes:

He has a meeting with the academic integrity committee next week. He's panicking. He has his original Spanish drafts and his translation process documented, but will they care??

If you've been through this or know someone who has, please share:

• What evidence helped?
• What did you say in the meeting?
• Did they understand the difference between translation and generation??

This could literally affect his visa status if he gets kicked out. So, you know, no pressure.

Thanks in advance. I'll pass along any advice.