AI Use Policy

Transparency in Artificial Intelligence
Effective Date: March 20, 2026 — Digital Dog Tech, a product line of iTeachAI LLC

1. Overview

Digital Dog Tech integrates artificial intelligence across its products — Blossom, College Buddy, and Open Rover — to enhance education. This policy explains how AI is used, what data it processes, and the safeguards we have in place.

We believe in AI transparency. Educators, parents, and students deserve to understand exactly how AI interacts with their data and what role it plays in the learning experience.

2. AI Tutor: Bubbles

2.1 Pedagogical Approach

Bubbles is designed to teach, not to do work for students. It uses Socratic questioning techniques, provides hints before answers, and encourages students to arrive at solutions themselves. Teachers can configure Bubbles’ behavior per assignment (e.g., hints only, explanations allowed, or fully guided).

2.2 Content Safety

Bubbles includes multiple layers of content filtering:

• Age-appropriate language and content controls based on the student’s grade level
• Refusal to engage with off-topic, inappropriate, or harmful queries
• Automatic flagging of concerning student inputs for teacher review
• Topic boundaries set by the teacher (e.g., only math for a math class)

3. AI-Assisted Lesson Planning

AI-generated lesson plans are always presented as drafts. Teachers review, edit, and approve all content before it reaches students. The AI serves as a starting point, not a replacement for professional judgment.

4. AI-Assisted Grading & Feedback

4.1 Teacher Approval Required

AI grading suggestions are never automatically applied. Every AI-suggested grade and comment must be reviewed and approved by the teacher before being visible to students. Teachers can modify, override, or reject any AI suggestion.

4.2 Transparency to Students

When AI-assisted grading is used, students and parents are informed that AI tools assisted in the grading process, and that the teacher reviewed and approved all final grades and feedback.

5. AI Models Used

Model	Provider	Use Case	Data Handling
Claude	Anthropic	Primary AI for tutoring, lesson planning, grading assistance, and content generation	Data sent via API is not used by Anthropic to train models. Anonymized before transmission.
Ollama (local models)	Self-hosted	On-premises processing for schools requiring data to stay on-site; lightweight tasks	All processing occurs locally. No data leaves the school’s infrastructure.

5.1 Why These Models

We selected Claude by Anthropic as our primary AI model because of Anthropic’s commitment to AI safety, their clear data usage policies (API data is not used for training), and the model’s strong performance in educational contexts.

We offer Ollama-based local models for schools and districts that require all data processing to occur on-premises, ensuring no student data ever leaves the school’s network.

6. Data Anonymization

Before any student data is sent to external AI models, we apply the following anonymization procedures:

• Name removal: Student names are replaced with anonymous identifiers (e.g., “Student A”).
• ID stripping: Student IDs, school names, and district identifiers are removed.
• Email removal: Any email addresses present in the content are redacted.
• Location scrubbing: Geographic identifiers (school address, city) are removed.
• Context isolation: Each AI request is processed independently; the AI model does not retain context between sessions or students.

7. No Training on Student Data

• Student interactions with Bubbles are not used to train any AI model (ours or third-party).
• Student work, assignments, and responses are not used as training data.
• Our agreement with Anthropic explicitly prohibits the use of API-submitted data for model training.
• Ollama-based local models are pre-trained open-source models; no fine-tuning on student data occurs.
• Anonymized, aggregated usage patterns (e.g., “80% of users found this feature helpful”) may be used to improve our product design, but never to train AI models.

8. Teacher Retains Control

AI is a tool that serves teachers, not a replacement for them. Teachers retain full control over:

• Whether AI is used: Teachers can enable or disable AI features per class, assignment, or student.
• What AI produces: All AI-generated content (lesson plans, grades, feedback) requires teacher review and approval.
• How Bubbles behaves: Teachers set the rules for AI tutoring (hint level, allowed topics, response style).
• Student access: Teachers control which students have access to AI features and can restrict access for individual students.
• Override authority: Teachers can override, modify, or reject any AI output at any time.

9. Opt-Out Options

We provide multiple levels of opt-out to respect the preferences of schools, teachers, parents, and students:

9.1 School/District Level

Administrators can disable AI features entirely for the entire school or district. All core platform functionality (classroom management, assignments, grading workflow) remains fully operational without AI.

9.2 Teacher Level

Individual teachers can disable AI features for their classes without affecting other teachers in the school.

9.3 Parent Level

Parents may request that their child not interact with AI features. The student retains full access to assignments, materials, and teacher feedback, but Bubbles and AI-generated content will not be available to them.

9.4 How to Opt Out

10. Accuracy Disclaimer

• AI outputs are suggestions and drafts, not authoritative educational content.
• Teachers are responsible for reviewing all AI-generated content before it is used with students.
• Students should be taught that AI is a helpful tool, but not always correct, and that critical thinking is essential.
• AI-generated grades are suggestions only and must be approved by the teacher.
• We continuously monitor and improve our AI systems to reduce errors, but perfection cannot be guaranteed.
• AI does not provide medical, legal, psychological, or crisis counseling advice. If a student’s input suggests they are in crisis, the system alerts the teacher.

11. AI Ethics & Safety

We are guided by the following principles in our use of AI:

• Transparency: We are open about where and how AI is used in our products.
• Human oversight: AI always operates under human supervision; it never makes autonomous decisions about students.
• Fairness: We monitor for and work to mitigate bias in AI outputs, including racial, gender, cultural, and socioeconomic biases.
• Privacy by design: Privacy protections are built into our AI pipeline, not added as an afterthought.
• Minimal data: We send only the minimum necessary data to AI models.
• Continuous improvement: We regularly review AI outputs, gather teacher feedback, and update our systems to improve safety and accuracy.

12. Updates to This Policy

As AI technology evolves, we may update this policy to reflect new capabilities, models, or safeguards. Material changes will be communicated via email to school administrators and posted on this page at least 30 days before taking effect.

13. Contact Information

If you have questions about how AI is used in our products, or to request additional technical details for your school’s review, please contact us: