My university uses 'Turnitin'. Does it flag Quillbot-ed text?

[SalmaWillis]

Okay, I'm asking for a friend. Obviously.

Grad school is kicking my butt and I'll admit—sometimes I use AI to help me rephrase things when I'm stuck. I'm not copying entire essays, but I might take a paragraph I wrote, run it through ChatGPT with "make this more academic" and then... tweak it.

But recently my university upgraded to Turnitin with AI detection. Apparently it can tell if something was written by AI, even if you paraphrased it?? Is that true??

Here's my specific question:

If I write a paragraph myself (my own ideas, my own words), then run it through Quillbot to "improve" the flow, will Turnitin flag that as AI? Because technically, Quillbot is AI, right?

I tested this myself with an old paper I wrote (100% human, got an A). I ran a few paragraphs through Quillbot on different settings and then put them into GPTZero (the free detector). Results:

• Fluency mode: 60% AI probability (yikes)
• Creative mode: 85% AI probability (double yikes)
• Formal mode: 40% AI probability (still yikes)

So apparently, yes, Quillbot DOES get detected at least sometimes. But GPTZero isn't Turnitin, so I don't know if that's accurate.

Questions for the brain trust:

1. Has anyone actually been caught by Turnitin for using Quillbot on their own writing?
2. Is there a "safe" way to use paraphrasing tools without triggering detectors?
3. If I manually edit the Quillbot output, does that help, or does the "AI pattern" still show up?

I'm not trying to cheat. I just... sometimes my brain doesn't words good.

Serious answers only please. I'm too old to get kicked out of grad school for being dumb about AI.