Our testing methodology
We tested 8 AI detectors using 500 text samples: 250 unmodified AI-generated texts from GPT-4o, Claude 3.5, and Gemini 1.5, and 250 human-written texts including academic papers, blog posts, and news articles. Each detector was run on every sample to measure true positive rate (AI correctly flagged) and false positive rate (human text incorrectly flagged).
| Sample type | Source | Count |
|---|---|---|
| AI-generated | GPT-4o, Claude 3.5, Gemini 1.5 | 250 |
| Human-written | Academic, blog, news | 250 |
| Total | All detectors run on every sample | 500 |
Test composition
Results: true positive rates
On unmodified AI text, these are the true positive rates (higher is better).
| Detector | True positive rate |
|---|---|
| Text Humanica Pro | 98.2% |
| Originality.ai | 97.3% |
| GPTZero | 96.1% |
| Turnitin | 94.8% |
| Copyleaks | 93.2% |
| Winston AI | 91.7% |
| ZeroGPT | 89.4% |
| Sapling | 88.9% |
True positive rate on unmodified AI text
The false positive problem
False positives are a serious problem because they flag human-written work as AI-generated, which can wrongly implicate students or writers. Lower is better here.
| Detector | False positive rate |
|---|---|
| Text Humanica | 3.1% |
| Originality.ai | 4.8% |
| Turnitin | 6.1% |
| GPTZero | 8.2% |
| ZeroGPT | 12.4% |
False positive rate on human-written text
Accuracy on humanized text
After running AI text through Text Humanica's Pro humanizer, here is how often each detector still caught it. Lower is better (means stronger bypass).
| Detector | Detected after humanization |
|---|---|
| GPTZero | 4.2% |
| Originality.ai | 5.8% |
| Turnitin | 7.1% |
| Copyleaks | 9.3% |
Detection rate after humanization
Which detector should you use?
For the most accurate detection of unmodified AI text, Text Humanica Pro or Originality.ai are the top choices. For academic use where false positives are a concern, Text Humanica's low 3.1% false positive rate makes it the safest choice. If you want the head-to-head detail, see our GPTZero, Turnitin, and Originality.ai comparisons.
- Highest accuracy: Text Humanica Pro (98.2%) and Originality.ai (97.3%).
- Lowest false positives: Text Humanica (3.1%).
- Strongest on humanized text: GPTZero detected only 4.2% of samples.
- For academic work, prioritize low false positive rate over raw accuracy.