AI Usage Policy

This policy describes how Rampant Strategy, LLC ("Rampant Strategy", "we", or "us") uses artificial intelligence within PeerDoc and PeerPoll (the "Services"), what information is sent to AI providers, and how that information is handled.

It is written so that instructors and institutions can reference it directly. Many universities now require that AI use affecting students be disclosed in a course syllabus. Rather than asking each instructor to describe our systems themselves, this policy is intended to be cited by reference, in the same way instructors commonly reference the posted policies of other academic tools. Suggested wording is provided at the end.

This policy supplements, and does not replace, our Privacy Policy and Terms & Conditions.

AI is used in three specific places in PeerDoc. It is not used anywhere else in the Services, and it is not used at all in PeerPoll.

1. AI Prompt activities (student-facing). An instructor may include an AI Prompt question in an assignment, allowing students to interact with an AI language model to research, test, or develop ideas. The full conversation, including student inputs and AI responses, is stored in PeerDoc and can be reviewed by the instructor and teaching assistants. These conversations are not private to the student.

2. Review assistance (instructor-facing). An instructor or teaching assistant may run an AI pass over written submissions for an assignment. It produces a preliminary quality ranking and a draft comment for each submission. These are advisory: they help staff decide which submissions to read first. Draft comments are never shown to students unless an instructor chooses to publish them.

3. AI-writing screening (instructor-facing). Where an institution has authorized it, submitted text may be checked by a third-party classifier that estimates whether writing was generated by AI. This feature is disabled by default and is enabled only with an institution's authorization. Results are reported to instructors as qualitative indicators, not as numeric certainty scores.

Before a submission is sent for AI-assisted review, the student's name, email address, and other identifying account information are removed and replaced with a numeric code that has no meaning outside PeerDoc. The code is pseudonymous rather than anonymous: PeerDoc can match results back to the student, but the AI provider cannot, and it does not receive the roster, the student's identity, or their institutional account details.

The submission itself is transmitted as written. Where a student includes their name or other identifying information within a response, within an uploaded document, or in that document's file properties, that information is transmitted with the submission. Uploaded files are stored under a generated identifier rather than the student's original filename, so the file name itself does not identify the student.

For AI Prompt activities, only the conversation is sent to the provider. No name, email address, or account identifier is attached, though anything a student chooses to type into the conversation is sent as part of it.

Rampant Strategy uses the following AI providers as sub-processors:

• OpenAI: used for AI Prompt activities and for review assistance. Specific models change over time as providers release and retire them; the models currently in use are available on request. OpenAI's data usage policy for its API states that submitted material is not used to train its models.
• Pangram: used for AI-writing screening, where an institution has authorized that feature. Screening is disabled unless enabled for an institution. Pangram's privacy policy states that submissions are not used to train its models and that it acts as a "school official" under FERPA for educational institutions.

We do not add AI providers without updating this policy. Institutions with an agreement requiring prior notification of new sub-processors will be notified in accordance with that agreement.

We use providers whose terms state that material submitted through their APIs is not used to train their models. Our requests are configured so that submitted material is not stored by the provider for later use. A provider may retain submissions for a limited period under its own published terms: OpenAI, for example, currently retains API logs for up to thirty days for abuse monitoring, after which they are deleted.

Material processed by AI remains subject to the retention and deletion terms described in our Privacy Policy. AI Prompt conversations, AI-generated rankings and comments, and AI-writing screening results are stored within PeerDoc as part of the course record, so that instructors and teaching assistants can review them.

AI output does not determine or calculate any portion of a student's grade. AI rankings are not converted into points, and no grade, academic finding, or other decision about a student is produced automatically by an AI system.

AI-assisted review is a reading aid. It orders submissions so that staff can prioritize their attention. Grades are assigned by the instructor or teaching assistant based on their own review of the work.

AI-writing screening does not produce a finding of academic dishonesty. Screening output is information for the instructor, who decides whether to pursue a concern under their institution's academic honesty process.

AI-writing screening is off by default and is enabled per instructor, with the institution's authorization. An instructor who has it enabled can still choose, on each run, whether screening takes place.

Review assistance is never automatic. It runs only when an instructor or teaching assistant starts it, for the assignment and questions they select.

AI Prompt activities appear only in assignments where an instructor has added them.

An institution that wishes to disable any AI feature for its courses should contact support@rampantstrategy.com.

Instructors are welcome to adapt the following, which is intended to be brief enough for a syllabus while pointing here for detail. Instructors should add their own statement of what AI use is permitted or prohibited for students in their course, which is a course policy rather than a platform one.

Changes that materially affect how student information is processed by AI will be communicated to affected institutions by email. Other changes will be posted here with an updated effective date. Because instructors reference this policy in course materials, we keep this page at a stable address.

Institutions evaluating these features can request our security documentation, accessibility conformance report, and current sub-processor and model details. Please contact support@rampantstrategy.com.