Last updated: 27-04-2026

This AI Transparency Policy explains how Drupto Consultants Pvt Ltd ("we," "us," or "our") uses Artificial Intelligence (AI) in DruptoQuiz (the "Platform"). It describes the AI models we employ, the purposes for which they are used, their limitations, and your rights regarding automated decisions. This policy is designed to comply with transparency obligations under the Digital Personal Data Protection Act, 2023 (DPDP Act) of India and other applicable enterprise-grade security standards.

1. Introduction

DruptoQuiz leverages AI to enhance the quiz creation, assessment, and proctoring experience. We believe in being transparent about our use of AI, including its capabilities, limitations, and the safeguards we have implemented. This policy outlines our commitment to responsible AI use and provides you with clear information about how AI impacts your interaction with the Platform.

2. AI Models Used

We integrate the following AI models and services into the Platform:

• Google Cloud Vertex AI (Gemini): We use the enterprise-grade Gemini models via Google Cloud Vertex AI for generating quiz questions, answers, and explanations. This managed infrastructure ensures that AI processing is isolated and secure, with project-specific IAM (Identity and Access Management) protections.
• MediaPipe Face Landmarker: A 3D computer vision model that detects 478 facial landmarks and movements during quiz sessions for proctoring purposes (e.g., detecting face presence and horizontal gaze direction).
• MediaPipe Object Detector (EfficientDet-Lite0): A lightweight, high-performance object detection model used to monitor the testing environment for unauthorized people or prohibited devices, such as cell phones.
• Custom Rule‑Based Algorithms: Non‑AI automated systems that enforce timing, navigation, and integrity rules during quiz attempts.

These models are accessed via secure APIs or run locally in your browser (for face detection) to protect your privacy.

3. AI‑Generated Content

AI is used to create quiz content (questions, multiple‑choice options, explanations) based on materials you provide or topics you specify. Please note:

• Accuracy Limitations: AI‑generated content may contain inaccuracies, biases, or errors. We recommend reviewing and editing generated quizzes before publishing.
• No Legal/Professional Advice: AI‑generated content is for educational and assessment purposes only and should not be relied upon as legal, medical, or professional advice.
• Human Oversight: You retain full control over the final content and are responsible for verifying its correctness and appropriateness.

4. Automated Decision Making

The Platform makes automated decisions in two key areas:

• Quiz Scoring: Multiple‑choice answers are automatically scored. Short‑answer responses may be evaluated by AI (Gemini) to provide a suggested score and feedback.
• Proctoring Flags: The system automatically flags potential integrity violations based on face detection, gaze analysis, and navigation patterns.

These automated decisions are not final or legally binding. They are subject to human review and appeal as described in Section 6.

5. Accuracy & Limitations

While we strive for accuracy, AI systems have inherent limitations:

• Error Rates: Quiz‑generation AI may produce factually incorrect or contextually inappropriate content in approximately 5‑10% of cases. Short‑answer assessment accuracy is estimated at 85‑90%.
• Proctoring False Positives: Face detection may incorrectly flag violations due to lighting changes, camera angles, or accessories (glasses, masks). The false‑positive rate is estimated at 3‑5%.
• Bias: AI models are trained on large datasets that may reflect societal biases. We implement mitigation measures (see Section 7), but bias may still affect generated content or assessments.
• Context Understanding: AI may not fully grasp nuanced or domain‑specific contexts, leading to suboptimal quiz questions or feedback.

We continuously monitor and improve our AI systems to reduce errors and biases.

6. Right to Opt-Out & Human Review

You have the right to opt-out of AI-powered features. For proctored quizzes, opting out may prevent participation if proctoring is a mandatory requirement for that specific assessment.

In accordance with our Terms and Conditions, you have the statutory right to request human review of any automated decision made by our AI proctoring or assessment systems. Such requests will be processed through our official grievance redressal channel.

7. Training Data & Zero-Training Guarantee

Training Data Sources: The AI models we use (Gemini, MediaPipe) are trained by Google on diverse public and licensed datasets. We utilize these as a managed service within Google Cloud Vertex AI.

Zero-Training Guarantee: We explicitly guarantee that **no user-provided data** (including prompts, text inputs, PDFs, or student responses) processed by Google Cloud Vertex AI on our Platform is used by Google to train, fine-tune, or improve any of its foundation models. Your data belongs to you and is processed in a secure, isolated environment.

DruptoQuiz Improvement: We may use anonymized, aggregated interaction data to improve our internal AI workflows, but we never use personally identifiable information for model development without your explicit opt‑in consent.

Bias Mitigation Measures: We implement the following to reduce bias and ensure fairness:

• Regular audits of AI‑generated content for stereotypical or discriminatory language
• Diversity checks in training data sampling where we have control
• User feedback loops to identify and correct biased outputs
• Transparency about known limitations in model documentation

8. User Rights Regarding AI Decisions

Under the DPDP Act, you have specific rights concerning automated processing:

• Right to Explanation: You can request a meaningful explanation of how an automated decision was reached, including the factors considered and the logic applied.
• Right to Correction: If you identify an error in the data used for an automated decision, you can request correction of that data.
• Right to Human Intervention: You have the right to request human review, express your point of view, and challenge the decision as described in Section 6.
• Right to Opt‑Out: For non‑essential automated decisions (e.g., content suggestions), you may opt out where technically feasible. Note that core platform functions (scoring, proctoring) require automation and cannot be opted out of while using the service.

To exercise these rights, contact us at saurabh.chandra@druptoconsultants.com.

9. Data Security & Privacy

AI processing respects your privacy:

• Local Processing: Face detection runs entirely in your browser; video/images are never transmitted to our servers.
• API Security/IAM: Calls to Vertex AI are authenticated via Google Cloud Identity and Access Management (IAM), ensuring project-wide isolation. Data transmission is governed by Google Cloud’s Data Processing Addendum (CDPA).
• Data Minimization: We only send the minimum necessary data (e.g., text excerpts for quiz generation) to AI services.
• Retention: AI‑generated content is stored as part of your quiz data and is subject to our standard data retention policies.

10. Changes to This Policy

We may update this AI Transparency Policy as our use of AI evolves or as legal requirements change. We will notify you of material changes by posting the updated policy on this page and updating the "Last updated" date. Continued use of the Platform after such changes constitutes acceptance of the revised policy.

11. Contact for AI‑Related Queries

If you have questions, concerns, or requests related to our use of AI, please contact:

AI Transparency Officer:
Saurabh Chandra
Email: saurabh.chandra@druptoconsultants.com

Grievances will be acknowledged within 24 business hours, and we aim to provide a formal resolution within 30 days of receipt, as per DPDP Act standards.

Phone: +91-6388063038 (available during business hours)
Address: JAIPURIA INNOVATION AND INCUBATION CENTRE, Hahnemann Rd, Vineet Khand, Gomti Nagar, Lucknow, Uttar Pradesh 226010

For general support, you may also contact support@druptoconsultants.com.

12. Licensing & Attribution

We leverage industry‑standard AI technologies from Google, ensuring high reliability and legal compliance:

• MediaPipe: All client‑side vision models are provided by Google and licensed under the Apache License 2.0. This permissive license explicitly allows for commercial use and modification.
• Vertex AI (Gemini): Our use of Generative AI is governed by the Google Cloud Platform (GCP) Service Terms and the official Data Processing Addendum (DPA), which provides the enterprise-grade data privacy and "Zero‑Training" guarantees cited in Section 7.