What is AI Detection?
Writable’s AI Detection (powered by Copyleaks) helps teachers determine whether a student’s writing was generated or assisted by AI.
To best assess writing authenticity, we recommend using Authorship Scores and Draft History with the AI Detection report to determine the likelihood of whether a student used AI to write a submission.
Accessing the Report
Knowing whether a submission was written by students or generated by AI is important for ensuring academic integrity. Our AI detector is trained to recognize human writing patterns and flags text as potential AI when it detects deviations from these patterns.
Log into Writable and open an assignment.
Click on a class section to open the Assignment Dashboard.
Click the "Students" view.
In the “AI Detection” column you will see a green document icon or a blue document icon.
🟩 Green icon: AI detection has run and either no AI was detected or there was insufficient student writing to evaluate. Hover over the icon to see which applies.
🟦 Blue icon: indicates that some AI use is suspected and there is a report available.Click on a blue icon to open the report.
❗️Note: Green icons will not open a report.
In the report, click any dark highlighted phrase to see details, including how often the phrase appears in AI-generated vs. human-written text.
Copyleaks uses large datasets and a complex AI algorithm to detect patterns in writing. These patterns are present in both human and AI-generated text, but differ in ways that allow for detection. You can learn more about how Copyleaks AI detection works here.
Note! Copyleaks works to minimize false positives through rigorous testing and ongoing model improvements, but they may still occur:
Writing Assistants: Features like Grammarly's genAI-driven content rewriting can cause text to be flagged as AI. Small adjustments for spelling and grammar typically do not trigger AI detection.
AI Humanizers: AI detection is trained on content altered by AI humanizers but is being continuously improved to ensure accuracy.