MIT and Cornell use TTA to improve AI text detection robustness
Understand this story
Researchers at MIT and Cornell University have proposed an AI text detection method based on test-time adaptation (TTA): after deployment, the detector can continuously adjust using new unlabeled data to keep up with changes in human and AI writing. According to MIT News, in experiments against AI text disguised through "humanization," the method achieved 93% recognition accuracy, compared with about 27% for commercial detectors; in scenarios simulating natural drift in human writing style, the method achieved 85% accuracy versus 73% for traditional methods. The researchers say the technique could be used in education, publishing, hiring, and for identifying phishing emails and SEO fraud, among other scenarios. The paper will be presented at the NeurIPS conference.
AI overview based on source reports · Updated 1 days ago
Coverage timeline
Follow the coverage to explore different perspectives.
- MIT News AIHelping AI text detectors keep up with changing conditions
MIT and Cornell researchers showed that AI text detectors can be fooled by disguised AI-generated text and by natural shifts in writing over time, and proposed using test-time adaptation, which lets a deployed detector adjust to new unlabeled data without retraining. In experiments, a commercial detector identified about 27 percent of disguised AI text while the TTA approach reached 93 percent accuracy, and under simulated natural language shift TTA was 85 percent accurate versus 73 percent for traditional methods. The researchers caution TTA may not always work as future LLMs produce more natural-sounding language and verifiable human text becomes harder to find.
Activity for this story
- Current comparable activity
- 3
- Peak comparable activity
- 5Nov 19 14:00
- Change in the past 24 hours
- –
Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.