Last Updated: June 1, 2026
This policy explains how Sukrat AI uses artificial intelligence (“AI”) to power the Sukrat platform, what data is and is not sent to AI providers, and what responsible use of Sukrat's AI features looks like.
1. What AI Does in Sukrat
Sukrat is a Complete Intelligent Tutoring System (ITS). AI is the engine behind its core educational capabilities:
- Adaptive assessment: Diagnosing your knowledge gaps by analyzing patterns in your answers across topics
- Mastery modeling: Building and updating a map of what you know and what you still need to master
- Calibrated practice generation: Creating practice questions pitched at exactly the right difficulty for your current mastery level
- Explanation and feedback: Generating exam-board-aware explanations, worked solutions, and targeted feedback on your answers
- AI Mentor and AI Chat: Answering curriculum questions conversationally, with context from your mastery state and current topic
AI assists your learning — it is not a replacement for your teacher, official course materials, or independent study.
2. Which AI Providers We Use
Sukrat uses multiple AI model providers to generate educational content:
- Anthropic (Claude API) — primary tutoring content generation and explanation
- OpenAI (GPT-4o API) — content generation and evaluation
- Google (Gemini API) — visual and diagram content generation
A full list of providers and what data they receive is on the Subprocessors page.
3. Our No-PII LLM Policy
We operate a strict policy to protect student privacy in AI processing:
- No student PII is ever sent to external AI providers. Names, email addresses, school affiliations, and any other personally identifiable information are stripped at the architectural level before prompts are constructed.
- Prompts contain only curriculum content: question text, syllabus reference, mark scheme context, and anonymized mastery state for the relevant topic.
- No student data is used to train AI models. This is enforced architecturally and required by contract with every AI provider Sukrat uses.
- AI providers may not use data they receive from Sukrat for any purpose other than returning a response to the specific request.
4. AI Accuracy and Limitations
AI-generated content can be incorrect, incomplete, or misleading. This is especially relevant for exam preparation:
- Always verify AI-generated worked solutions and explanations against official exam board mark schemes and examiner reports
- AI-generated practice questions are calibrated for learning but are not guaranteed to match the exact difficulty, wording, or style of official exam questions
- If an AI explanation contradicts your teacher's guidance, ask your teacher — official sources take precedence
- Sukrat does not guarantee exam results, grades, or specific academic outcomes. Performance depends on your preparation effort, study time, and many factors outside the platform.
If you encounter content that appears incorrect, please report it using the feedback tool in the platform. We use these reports to improve accuracy.
5. AI and Academic Integrity
Sukrat's AI is designed as a learning tool, not a shortcut. Using it responsibly means:
- Engage, don't copy. Read AI explanations to build your understanding, not to copy answers verbatim into your work.
- Attempt first. Try practice questions yourself before asking the AI for a worked solution. The attempt is where the learning happens.
- Follow your school's rules. AI-generated content must not be submitted as your own work in any assessed context where AI assistance is prohibited. Your school's academic integrity policy and AI use policy govern how you may use Sukrat in assessed contexts. Where your school's policy is stricter, it takes precedence.
For the full rules on acceptable use of Sukrat's AI, see our Acceptable Use Policy.
6. AI Safety and Safeguards
Sukrat implements safeguards to reduce harmful or off-topic AI outputs:
- AI responses are scoped to curriculum content — the system prompt restricts the AI to its educational purpose
- AI outputs are evaluated using Langfuse for quality and safety before deployment changes
- User feedback on incorrect content is reviewed by the Sukrat team
No AI system is perfect. If you encounter output that is harmful, offensive, or seriously incorrect, report it to legal@sukrat.ai.
7. How AI Data Processing Relates to Privacy
AI processing is governed by our Privacy Policy. Key points:
- Session transcripts (AI interaction logs) are retained for 90 days and then deleted
- Sukrat does not sell your educational data or AI interaction history
- Sukrat does not use your personal data or session transcripts to train AI models
8. Changes to This Policy
We may update this policy as our AI systems evolve. Material changes will be communicated via email or in-product notification.
9. Contact
Questions about AI usage on Sukrat? Contact us at legal@sukrat.ai.
Last Modified: June 1, 2026
Policy Version: 2.0