Implement
AI Policy
with Confidence
for Educational Institutions
No. Rumi is not an AI detector. In fact, we are against the use of AI detectors due to their ineffectiveness and inaccuracy. We promote integrating responsible and manageable AI policies into classrooms. Click here to learn more on why schools should not rely on AI Detectors.
Yes! Instructors have the freedom to select which AI policy they want to set to each assignment. They can also manage and set the allowed prompts and limitations. Instructors can select from:
No AI: Fully restricted AI use for the assignment.
Partial AI: Only allowing a limited selection of prompts to be used, such as ''Fix Grammar and Spelling'', ''Improve Sentence Structure'' and more. Prompts can also be personalized by instructors.
Full AI: Unrestricted use of AI for the assignment.
Rumi saves student's AI usage history for instructor to later review.
Rumi collects only draft data, related analytics, and AI prompt usage, similar to Google Docs. However, unlike Google Docs, Rumi shares all draft, analytics and AI usage data directly with instructors. Rumi does not lock browsers, monitor other tabs, record screens, or request any additional access.
AI Engines utilized by Rumi are licensed with agreements that prohibit training on student-generated content.
Rumi offers instructors two distinct methods to identify "transcribing":
1. For transcribed papers, two key metrics we monitor—typing time and the number of revisions (such as deletions and edits)—stand out from the other metrics in the class, showing a significantly lower frequency in comparison to authentic essays.
2. The revision history for transcribed essays generally follows a straight, linear progression, unlike authentic essays where students tend to revise and rework their content in a more iterative, back-and-forth manner.
No. Rumi retains data for the duration necessary to comply with FERPA requirements. However, student-written content is not, and will not be, used for comparison with other students' materials. We do not consider this approach viable because:
A) It monetizes student work without most students being aware of it.
B) With the rise of AI, comparing stored text to determine originality is not very effective, as AI-generated text is uniquely created each time.
For more details, please refer to the "Data Ownership" section of our terms of policy.
Rumi AI is currently connected to Claude 3.5, Gemini and GPT-4o. Students have the option to use one or multiple AI engines for an assignment if the instructor allows it.
AI Engines utilized by Rumi are licensed with agreements that prohibit training on student-generated content.
AI human-like typing tools and simulators do not work on Rumi. Any attempt to override Rumi's functions will be automatically blocked.
Rumi utilizes a search API to compare text with publicly available content on the Internet. The score is based on the similarity between the document's content and information found online, such as on Wikipedia. Student-written content is not, and will not be, used for comparison with other students' materials
Yes, Rumi works with any language.