The August 27, 2025 Release Explained
On August 27, 2025, Turnitin released its most aggressive AI detection update yet. Unlike earlier models that mainly flagged text from ChatGPT, Gemini, or Claude, this update was built to detect AI humanizers. Turnitin now claims it can expose tools that “humanize” AI writing.
The reality? TwainGPT still bypasses Turnitin. Even after this update, tests show that TwainGPT still produces human results that bypass AI detection.
What Changed in the Update
Turnitin’s new release adds AI bypasser detection to its existing AI writing and paraphrasing models.
Now:
- Cyan highlights = AI-generated or bypassed text.
- Purple highlights = AI-generated + paraphrased text.
- Results are bundled into the AI report with a percentage score.
Turnitin’s pitch: no bypasser can escape.
The Ongoing Problem
Even with new models, Turnitin still struggles:
- False positives — genuine human writing flagged as AI.
- Low confidence — scores from 1-19% are marked as “*”.
In other words, the update may sound promising, but it’s far from perfect.
Why TwainGPT Still Works
TwainGPT avoids the predictable patterns AI detectors look for. Instead of relying on obvious tricks like random spacing, character swaps, etc., TwainGPT takes a smarter approach:
- Restructures sentences for natural, human-like flow.
- Expands vocabulary to break AI word patterns.
- Adds human-like characteristics.
Test Example (Sept. 3, 2025)
- Raw GPT-5 AI essay detected at 100% AI.
- After TwainGPT → 0% AI, detected as fully human.
Final Thoughts
Turnitin’s August 2025 update was successful at detecting most AI humanizer tools, but not TwainGPT.
While other humanizers now trigger Turnitin’s new detection flags, TwainGPT continues to stay ahead, producing undetectable, natural writing that bypasses every AI detector.
For students, researchers, and professionals, TwainGPT remains the #1 effective way to bypass Turnitin in 2025.