Copyleaks
Copyleaks is CCP's plaigiarism and AI detection tool. It can be enabled for assignments in Canvas and will automatically scan submissions against a database of print and digital literature to detect plaigiarism and analyze the submission for tells that indicate the use of generative AI.
Note: Copyleaks is not compatible with discussions or quizzes but, as a workaround, their Teacher Scan Tool allows anything to be uploaded and scanned.
Getting Started
These three videos will give you a crash course in Copyleaks, covering the creation of a Copyleaks assignment and interpreting the Similarity Reports it generates. (~8 min)
Creating an assignment with Copyleaks (~3 min)
Viewing student submissions & similarity reports (~1 min)
Interpreting the Similarity Report (~3.5 min)
How do I know I can trust Copyleaks?
Copyleaks publishes their methodology on their website, along with studies on it's accuracy using the different scan sensitivity levels provided.
Copyleaks Teacher Scan Tool
The Teacher Scan Tool can be added to a module in your course and will allow you to upload any text or file for similiarity detection. While versatile, this version of Copyleaks will only run on the material you manually upload to it. One or multiple files can be uploaded at once. This way, the tool can be used to scan discussion posts or essays submitted as part of a quiz.
Learn more about the Teacher Scan Tool on The Professional Learning Hub. (Note that you must be logged into Canvas for this link to work.)
Responding to a Positive Response
Use Your AI Policy
It's always best to have an AI policy in your syllabus from the beginning. That way if a student does use AI in a way you don't allow, you have that agreement in the syllabus to refer to.
Understand that the Copyleaks report is just one data point
While rare, Copyleaks occasionally does make mistakes. Always look for other signs and evidence that AI was used.
Look at the student's writing history/analytics
Does the style of the submission match the student's previous work? Is the style consistent? Is there a high rate of phrases that are commonly used by AI? Is the writing or vocabulary appropriate for the student's level? Refer to previous reports to determine how often the student submits content that Copyleaks flags.
Talk with the student
For instance, call them to your office and have them answer questions about or give a synopsis of the submission. Did they cover the same items that were in the submission? If some AI is allowed in your class, have them provide the method of research and any prompts they used for the assignment.
Consider what AI means for you
Because Copyleaks anaylzes the document submitted, it doesn't always know where the AI content may have come from. For instance, while Grammarly is widely accepted as a supporting tool for writing, some of it's features do get flagged in Copyleaks reports as AI. If you are allowing similar tools, it may be a false positive due to what's allowed and what isn't.
A portion of the text has been flagged as AI, what does that mean?
- AI generated content that Copyleaks flags will be highlighted in purple. An overall percentage illustrating how much AI content makes up the document is displayed at the top of the page. Below this score is a breakdown of AI Phrases, which shows the relative frequency of phrase usage by AI versus humans.
- Even if text is highlighted, it does not mean every single word is AI generated, as Copyleaks analyzes text in overlapping sections. This could explain why things like headers, references and titles may be highlighted.
- When AI writes a sentence, it probes all of its pre-training data to output a statistically generated sentence, which often does not resemble the patterns of human writing. It becomes more apparent when analyzed against a vast corpus of human writing.
- AI Detectors can offer a lot of insight and data to encourage essential conversations in classrooms and boardrooms alike to determine the rules and regulations around AI. When AI content is detected, take the time to investigate further. Again, the data provided by AI detectors should be used to inform the situation, create a learning opportunity, and align on expectations.
What do the different colored flags next to a Copyleaks assignment mean?
Once a Copyleaks scan finishes, a colored flag will appear next to the student's grade and submission. The color is a rough estimation of the originality score they recieved.
- Blue: No matching text
- Green: 1-24%
- Yellow: 25-49%
- Orange: 50-74%
- Red: 75-100%