Skip to main content

Which AI models does CoGrader use, and what data do they see?

The AI providers CoGrader uses, what they see, and how student data is protected.

Written by Gabriel Chi

CoGrader uses AI models from Google, OpenAI, Microsoft and Anthropic to grade student work and write feedback. They receive the assignment, rubric and student answer, without student names, and they can't use that data to train their models.

Our AI providers

  • Google LLC: Gemini, through Google Cloud Vertex AI and Google AI Studio. This is our primary provider.

  • OpenAI OpCo, LLC: the OpenAI API.

  • Microsoft Corporation: Azure AI Foundry.

  • Anthropic, PBC: Claude, through the Anthropic API.

What the AI sees

  • The assignment prompt and any source documents

  • Your rubric or answer key

  • The student's answer

Student names and other identifiers are removed before work is sent to an AI provider.

How the data is protected

  • No training: customer data, including student work, prompts, rubrics and AI feedback, is not used to train AI models. Our contracts bar providers from using it to train or improve their models.

  • Short retention by providers: a provider may keep a temporary copy for up to 30 days for abuse and safety monitoring, then delete it.

  • Storage: CoGrader stores data in the United States on Google Cloud.

  • Independent audit: CoGrader has a SOC 2 Type 1 attestation.

Questions

Email privacy@cograder.com or read the full Privacy Policy.

Related

Did this answer your question?