Machine-Generated Text Detection on Reuters (Claude)
85.2TPR @ FPR=1%DNA-GPT-M
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DNA-GPT-MTraining Generator=Dolly, Calibration=Markov-informed2026.02 | 85.2 | 99.47 | |
| DNA-GPTTraining Generator=Dolly, Calibration=None2026.02 | 76.76 | 99.21 | |
| DNA-GPT-MTraining Data=ChatGPT, Variant=Markov-informed score calibration2026.02 | 61.91 | 98.78 | |
| Likelihood-MTraining Generator=Dolly, Calibration=Markov-informed2026.02 | 31.96 | 90.57 | |
| Log-Rank-MTraining Generator=Dolly, Calibration=Markov-informed2026.02 | 31.78 | 89.98 | |
| Log-Rank-MTraining Data=ChatGPT, Variant=Markov-informed score calibration2026.02 | 29.73 | 89.55 | |
| Likelihood-MTraining Source=ChatGPT-turbo2026.02 | 29.6 | 90.83 | |
| Log-Rank-MTraining Source=ChatGPT-turbo2026.02 | 27.29 | 89.02 | |
| Likelihood-MTraining Data=ChatGPT, Variant=Markov-informed score calibration2026.02 | 26.71 | 90.39 | |
| Log-RankTraining Data=ChatGPT2026.02 | 24.58 | 85.68 | |
| Log-RankTraining Source=ChatGPT-turbo2026.02 | 24.58 | 85.68 | |
| DNA-GPTTraining Data=ChatGPT2026.02 | 21.96 | 97.21 | |
| LikelihoodTraining Data=ChatGPT2026.02 | 19.42 | 86.46 | |
| LikelihoodTraining Source=ChatGPT-turbo2026.02 | 19.42 | 86.46 | |
| LikelihoodTraining Generator=Dolly, Calibration=None2026.02 | 19.38 | 86.45 | |
| Entropy-MTraining Data=ChatGPT, Variant=Markov-informed score calibration2026.02 | 18.09 | 83.33 | |
| Entropy-MTraining Source=ChatGPT-turbo2026.02 | 17.42 | 83.14 | |
| Entropy-MTraining Generator=Dolly, Calibration=Markov-informed2026.02 | 9.24 | 56.78 | |
| DetectGPT-MTraining Data=ChatGPT, Variant=Markov-informed score calibration2026.02 | 8.27 | 75.78 | |
| DetectGPT-MTraining Generator=Dolly, Calibration=Markov-informed2026.02 | 6.27 | 57.53 | |
| Log-RankTraining Generator=Dolly, Calibration=None2026.02 | 4.93 | 28.3 | |
| EntropyTraining Data=ChatGPT2026.02 | 4.36 | 76.24 | |
| EntropyTraining Source=ChatGPT-turbo2026.02 | 4.36 | 76.24 | |
| DetectGPT-MTraining Source=ChatGPT-turbo2026.02 | 4.18 | 73.92 | |
| DNA-GPT-MTraining Source=ChatGPT-turbo2026.02 | 3.78 | 67.14 | |
| DNA-GPTTraining Source=ChatGPT-turbo2026.02 | 1.96 | 43.56 | |
| FastGPTTraining Data=ChatGPT2026.02 | 1.42 | 58.69 | |
| FastGPTTraining Source=ChatGPT-turbo2026.02 | 1.42 | 58.69 | |
| FastGPT-MTraining Generator=Dolly, Calibration=Markov-informed2026.02 | 1.38 | 67.04 | |
| FastGPT-MTraining Data=ChatGPT, Variant=Markov-informed score calibration2026.02 | 0.89 | 66.72 | |
| DetectGPTTraining Source=ChatGPT-turbo2026.02 | 0.62 | 41.99 | |
| FastGPT-MTraining Source=ChatGPT-turbo2026.02 | 0.62 | 50.18 | |
| DetectGPTTraining Data=ChatGPT2026.02 | 0.44 | 58.01 | |
| DetectGPTTraining Generator=Dolly, Calibration=None2026.02 | 0.44 | 58.01 | |
| FastGPTTraining Generator=Dolly, Calibration=None2026.02 | 0.09 | 41.31 | |
| EntropyTraining Generator=Dolly, Calibration=None2026.02 | 0 | 23.76 |