PoliciesAI Policy

Human-first AI

How we use artificial intelligence at Optimem and how it affects your data.

Updated Mar 3, 2026

Optimem offers some optional AI-powered features, like chat tutors and content tools, to help you learn. These features are clearly labelled and optional. If you're not actively using or opted-in to an AI feature, no data is being sent to any AI provider. This policy explains how AI works when you do use it, what happens with your data, and what we expect when you create content with AI.

Our principles

AI should enhance learning, not replace thinking. Every AI feature in Optimem follows these rules:

  • Clearly labeled. You'll always know when AI is involved.
  • Optional. You can disable individual AI features or turn off AI entirely in Settings. Nothing activates without your choice.
  • Transparent. You'll know what data is sent and to whom.
  • Human-first. Our AI is designed to upgrade your learning, not replace it. We use professional, crowd-sourced, and wiki-based verification and grounding to keep AI output accurate.
  • Privacy-respecting. We never save personal message data and minimize what we share with AI providers.

How we use AI

In the product

  • AI chat tutors. Conversational assistants for language practice, material review, and interactive learning.
  • Content generation assistance. Tools to help create, review, translate, or improve flashcard content.
  • Content review and quality. AI-assisted checks for accuracy and completeness.

All user-facing AI features are clearly labeled.

In development

Our team uses AI-assisted coding and design tools. We're a small, self-funded team, and these tools help us build faster. Any content we publish that's primarily AI-generated is clearly labeled.

AI providers

When you use an AI feature, your input is sent to one of these providers to generate a response.

Anthropic: Not used for training. Up to 30 days retention (trust & safety), then deleted. Privacy policy
OpenAI: Not used for training. Up to 30 days retention (abuse monitoring), then deleted. Privacy policy
AssemblyAI
AssemblyAI: Speech-to-text for voice input. We never save your recordings. AssemblyAI may retain audio temporarily per their terms, including for model improvement. Privacy policy

We also design our integrations to minimize what's transmitted, using anonymization, truncation, and context windowing where possible. We don't separately save your chat messages or audio recordings.

What data goes to AI providers

Sent: Chat messages you send to AI tutors, content you submit for AI-assisted generation or review, relevant context needed for a useful response (like language or topic), and some learning context (like which concepts you find difficult) to personalize your experience.

Not sent: Your email address, account credentials, payment information, or analytics data.

Retention: We don't store AI conversation contents, messages are processed in real time and discarded from our servers. Our providers may retain message data for up to 30 days for trust and safety, then delete it.

You can disable AI features entirely in Settings. This won't affect core study and flashcard features.

Accuracy and limitations

AI is a useful tool but not a perfect one. AI-generated content may contain errors, inaccuracies, biases, or hallucinations. It's for educational assistance -- not professional, medical, legal, financial, or other expert advice. You're responsible for reviewing and verifying AI-generated content before using or sharing it.

We use community grounding, wiki-based verification, and human review to improve quality, but no system is infallible.

AI in user-created content

You're welcome to use AI to help create learning packs. If you do:

  • Disclose it. Label AI-generated or AI-assisted content honestly. The Optimem editor automatically labels content created with its built-in AI tools.
  • Add value. Review, edit, and improve AI output meaningfully. Low-effort, bulk-generated content with no human oversight isn't allowed.
  • Verify accuracy. You're responsible for fact-checking AI output before sharing.
  • Follow the rules. AI-generated content must comply with our Content Policy and Terms of Service.

Other policies