Official Legal Policy

NeuraMach AI Studio Private Limited - AI Usage Policy

Effective Date: June 11, 2026Last Updated: June 11, 2026

ScoreVedaa CAT is a product owned and operated by NEURAMACH AI STUDIO PRIVATE LIMITED, a private limited company incorporated under the laws of India, bearing Corporate Identification Number U62099PN2025PTC245340 and operating under the brand name NeuraMach.ai.

Registered Office:

Flat No. 201, Building 1, Wing 3, The Crown Greens, Plot 17, Infotech Park, Hinjawadi, Pune, Maharashtra 411057, India.

Principal Business and Correspondence Office:

3rd Floor, Cabin No. 7, Quick Office, 301, 45 Baner Road, above Atithi Restaurant, Veerbhadra Nagar, Baner, Pune, Maharashtra 411045, India.

In this AI Usage Policy, references to “ScoreVedaa”, “ScoreVedaa CAT”, “NeuraMach.ai”, “Company”, “we”, “us” or “our” mean NEURAMACH AI STUDIO PRIVATE LIMITED, unless the context requires otherwise.

This AI Usage Policy explains:

  • How artificial intelligence and automated systems may be used through ScoreVedaa CAT;
  • The purposes and limitations of AI Features;
  • How AI Outputs should and should not be used;
  • How User information may be processed through AI systems;
  • The distinction between AI inference, Account-Level Personalisation and Reusable Model development;
  • The conditions governing internal and provider-side model training;
  • The responsibilities of Users, Institutes and Faculty Users;
  • The safeguards applicable to AI providers, human review and automated analysis;
  • Prohibited uses of AI Features;
  • How inaccurate, harmful or inappropriate Outputs may be reported; and
  • The circumstances in which access to AI Features may be reviewed, restricted or suspended.

Please read this Policy together with the:

  • ScoreVedaa CAT Terms and Conditions;
  • ScoreVedaa CAT Privacy Policy;
  • ScoreVedaa CAT Cookie Policy;
  • Subscription, Cancellation and Refund Policy;
  • Model Improvement Notice, where applicable;
  • Provider Training Notice, where applicable;
  • Institute Agreement, where applicable;
  • Applicable consent records; and
  • Any feature-specific notice displayed before or during use of an AI Feature.

A. OVERVIEW, SCOPE AND DEFINITIONS

1. Introduction and Purpose

ScoreVedaa CAT may use:

  • Artificial intelligence;
  • Machine learning;
  • Generative AI;
  • Natural-language processing;
  • Statistical analysis;
  • Automated rules;
  • Retrieval systems;
  • Recommendation systems;
  • Classification systems; and
  • Human-designed educational frameworks

to support CAT preparation and related Platform functions.

AI may assist an Adult User in understanding:

  • Assessment performance;
  • Demonstrated strengths;
  • Areas requiring further preparation;
  • Possible causes of errors;
  • Time-allocation patterns;
  • Question-selection patterns;
  • Topic priorities;
  • Difficulty Coverage;
  • Potential next steps;
  • Preparation goals;
  • Relevant practice activities; and
  • Explanations relating to CAT questions and concepts.

AI is used primarily as an educational-support and operational tool. AI is not intended to:

  • Replace official CAT notifications;
  • Replace official answer keys;
  • Guarantee a score or percentile;
  • Guarantee an interview call or admission;
  • Replace qualified teachers in every circumstance;
  • Make an official academic or admission decision;
  • Diagnose a medical, psychological or cognitive condition;
  • Determine intelligence or personal worth;
  • Provide medical, legal, investment, tax or financial advice;
  • Make employment, lending, insurance or Government-benefit decisions; or
  • Determine a User’s legal rights.

Users must review AI Outputs critically and must not assume that an Output is correct merely because it is generated, displayed or recommended by ScoreVedaa CAT.

2. Scope of this Policy

This Policy applies to AI and Automated Processing used through:

  • The ScoreVedaa CAT website;
  • Student dashboards;
  • Institute and Faculty dashboards;
  • PRGNA;
  • Assessment-generation and assembly functions;
  • Assessment-analysis functions;
  • The Exam Simulator;
  • Topic Classification;
  • Difficulty Coverage;
  • Confidence, Readiness and Coverage indicators;
  • Goal and Action-Plan features;
  • Personalised Recommendations;
  • Question-selection systems;
  • User-upload analysis;
  • Error and root-cause analysis;
  • Institute-facing educational insights;
  • AI-assisted support tools;
  • Content-moderation systems;
  • Security and fraud-detection systems;
  • Trial, beta and demonstration features;
  • Web and mobile applications; and
  • Any future ScoreVedaa CAT feature identified as AI-assisted or automated.

This Policy applies to:

  • Adult Visitors using an AI Feature;
  • Registered Adult Students;
  • Individual subscribers;
  • Institute-sponsored Adult Students;
  • Institute Administrators;
  • Faculty Users;
  • Adult mentors and counsellors;
  • Authorised support personnel; and
  • Other authorised Adult Users.

This Policy does not govern:

  • Independent AI tools used outside ScoreVedaa CAT;
  • Processing independently undertaken by an Institute outside the Platform;
  • AI systems used by an employer, college or other organisation without ScoreVedaa’s direction;
  • A separately identified ScoreVedaa JEE, NEET, school-board or other product; or
  • Third-party websites and services governed by their own terms and policies.

3. Eligibility and Age Requirement

ScoreVedaa CAT is a CAT-exam preparation platform intended for adult learners (18+) because its target examination and use case — CAT and MBA admissions — apply only to adults. This restriction is unrelated to the nature of the content, which is educational. The term 'Adult' below is used only in this defined, age-eligibility sense. ScoreVedaa CAT is intended only for individuals who have completed eighteen years of age and are legally capable of entering into a binding agreement. An individual below eighteen years of age must not:

  • Register an Account;
  • Use PRGNA;
  • Take an Assessment;
  • Upload User Content;
  • Use an Account-based AI Feature;
  • Purchase a Subscription;
  • Participate in model-improvement programmes;
  • Use another person’s Account; or
  • Circumvent an age-assurance measure.

A parent, guardian, Institute, teacher or other person may not activate a personal ScoreVedaa CAT Student Account for an individual below eighteen years of age.

If we reasonably determine that an Account may be used by an individual below eighteen, we may promptly take one or more proportionate measures, including restricting the Account, suspending relevant AI Features, requesting verification, stopping age-restricted Processing, cancelling unauthorised access and deleting, securely isolating or lawfully retaining associated information as appropriate.

An Adult User must not upload a minor’s Personal Data or request AI processing concerning a minor without lawful authority and a legitimate purpose.

Age assurance is not based on a self-declaration alone. We collect and validate date of birth at registration, apply additional verification where information is inconsistent or account sharing is suspected, and require Institutes to warrant that every student on a submitted roster is an adult.

4. Relationship With Other Documents

This Policy governs the operation, limitations and permitted use of AI Features and AI Outputs. Where provisions conflict:

  • Mandatory Applicable Law prevails.
  • A signed Institute Agreement or data-processing agreement governs the relevant institutional relationship.
  • A specific Model Improvement Notice and corresponding Training Consent govern optional internal identifiable-data Reusable Model development.
  • A specific Provider Training Notice and corresponding separate Consent govern any approved provider-side model development.
  • The Privacy Policy governs general Personal Data Processing.
  • The Cookie Policy governs Cookies and Similar Technologies.
  • This Policy governs AI Features, AI Outputs, AI limitations and acceptable AI use.
  • The Terms and Conditions govern general access to and use of ScoreVedaa CAT.

General acceptance of the Terms or acknowledgement of the Privacy Policy does not constitute:

  • Internal Training Consent;
  • Provider-side model-training Consent;
  • Consent to unrelated human review;
  • Consent to optional Cookies;
  • Consent to cross-product model development; or
  • Consent to another ScoreVedaa product.

Using an AI Feature to obtain an immediate response does not, by itself, constitute Consent to use the relevant information for Reusable Model development.

5. Definitions

For this Policy:

  • “Account-Level Personalisation” means Processing undertaken to customise or adapt the Services for a particular User or Account without training a Reusable Model for general use across other Users.
  • “Adult” means an individual who has completed eighteen years of age or such higher age as may be required to enter into a binding agreement under Applicable Law.
  • “AI” means artificial intelligence, machine learning, generative AI, natural-language processing, statistical models, automated decision rules, retrieval systems and related computational systems.
  • “AI Feature” means a ScoreVedaa CAT function that uses AI or Automated Processing to generate, classify, recommend, analyse, explain, rank, predict, detect, personalise or assist.
  • “AI Incident” means an event involving an AI Feature that may create material risk, including: exposure of Private User Content; unauthorised model-training use; security compromise; repeated materially incorrect Outputs; harmful or unlawful Outputs; material discriminatory behaviour; prompt injection; model manipulation; data leakage; provider breach; copyright reproduction; failure of a safety control; or another serious malfunction.
  • “AI Output” means a question, answer, explanation, recommendation, classification, estimate, plan, summary, analysis, label, score-related indicator, priority ranking, prediction or other output generated or assisted by an AI Feature.
  • “AI System” means a Company-controlled or third-party model, software component, retrieval system, automated rule, statistical method or technical system supporting an AI Feature.
  • “Applicable Law” means any law, rule, regulation, order, direction or legally binding requirement applicable to the Company, an AI Feature, a User, an Institute or the relevant Processing.
  • “Assessment” means any diagnostic, topic, chapter, sectional, mock, practice, foundation, maintenance, stabilisation, challenge or other test offered through ScoreVedaa CAT.
  • “Automated Processing” means Processing performed wholly or partly using AI, algorithms, statistical systems, automated rules or related technologies.
  • “Data Processor” means a person or entity that processes Personal Data on behalf of and under the documented instructions of a Data Fiduciary, or as otherwise defined under Applicable Law.
  • “Difficulty Coverage” means an internal estimate of the difficulty level or range at which a User has demonstrated a specified level of performance based on available evidence.
  • “Faculty User” means an Adult teacher, mentor, counsellor or similar individual authorised by an Institute.
  • “High-Impact Decision” means a decision materially affecting an individual’s access to education, employment, lending, insurance, Government benefits, healthcare, legal rights or another consequential opportunity.
  • “Institute” means a coaching institution, educational organisation, training provider, college, company or other organisation that purchases, sponsors, administers or facilitates Adult access to ScoreVedaa CAT.
  • “Institute-Linked Account” means an Account connected to an Institute, batch, programme or institutional Subscription through an authorised process.
  • “Material AI Feature” means an AI Feature whose Outputs, data use, visibility, scale, novelty or potential effect creates a level of risk requiring enhanced governance, testing, notice or review.
  • “Model Improvement Notice” means a separate notice describing optional internal Reusable Model development, including relevant data categories, purposes, historical-data use, human review, approved Data Processors, retention and withdrawal.
  • “Performance Data” means information concerning Assessments, answers, attempts, scores, timing, classifications, preparation patterns, goals and related activity.
  • “Personal Data” means data about an individual who is identifiable by or in relation to that data, or as otherwise defined under Applicable Law.
  • “Personalised Recommendation” means a suggestion concerning preparation priorities, Assessments, topics, activities, revision, milestones or next actions generated wholly or partly using available User data.
  • “Private User Content” means non-public prompts, conversations, uploads, Assessment answers, study notes, support communications and other User Content not intentionally made public by the User to the general internet. Sharing information with a specific Faculty User or within a closed Institute programme does not make it public.
  • “PRGNA” means ScoreVedaa CAT’s AI-supported learning and doubt-assistance feature, or a successor feature.
  • “Provisional Output” means an Output based on limited, incomplete, inconsistent, outdated or insufficient evidence and which may change after further activity or review.
  • “Provider Training Notice” means a separate notice describing proposed use of Personal Data by a third-party provider for the provider’s independently controlled model training or product improvement.
  • “Reusable Model” means an AI, machine-learning, statistical or automated model, system or component trained, fine-tuned, tested, evaluated, validated, calibrated, benchmarked or improved using data so that the resulting learning may be used beyond the immediate User-requested transaction, session or Account-Level Personalisation.
  • “ScoreVedaa Content” means questions, explanations, software, workflows, interfaces, reports, analytics, models and other Content owned, licensed, developed, commissioned or lawfully made available by the Company.
  • “Synthetic Content” means information artificially or algorithmically created, generated, modified or altered using a computer resource, including content treated as synthetically generated information under Applicable Law.
  • “Training Consent” means separate affirmative Consent through which a User agrees to specified optional internal Reusable Model development involving identified categories of Private User Content or identifiable Performance Data. Training Consent is not created by: acceptance of the Terms; acknowledgement of the Privacy Policy; use of an AI Feature; silence; inactivity; continued use; failure to opt out; a pre-selected checkbox; or a pre-enabled setting.
  • “User Content” means a prompt, question, screenshot, answer, note, document, image, file, message, feedback or other material submitted, uploaded or transmitted by a User.
  • “User”, “you” or “your” means an Adult individual or authorised organisation using or interacting with ScoreVedaa CAT.

Capitalised terms not separately defined in this Policy have the meanings assigned in the Terms and Conditions or Privacy Policy.

B. RESPONSIBLE AI PRINCIPLES AND OPERATION

6. Nature of AI-Assisted Services

AI Features may be entirely automated, rule-based, statistical, generative, retrieval-assisted, human-authored, human-reviewed, or produced through a hybrid workflow. Not every ScoreVedaa CAT output is generated by AI. Not every AI Output is reviewed by a human before being displayed. The Company will not represent that an Output was reviewed, verified or approved by a human unless such review actually occurred.

AI Features may use multiple systems or providers in sequence. For example, an AI Feature may extract text from an upload, retrieve relevant ScoreVedaa Content, apply rules, generate a draft response, perform automated safety checks, and display the resulting Output. The use of more than one system does not guarantee accuracy.

7. Principles Governing AI Use

The Company will apply safeguards proportionate to the nature, scale and risk of the relevant AI Feature. Its responsible-AI approach is guided by the following principles.

7.1 Lawfulness and purpose limitation

AI-related Processing will be undertaken only for lawful, disclosed and authorised purposes. Information will not be repurposed for an incompatible AI use without appropriate notice and any required Consent or other lawful ground.

7.2 Safety and reliability

Material AI Features will be designed, tested, monitored and operated with safeguards proportionate to foreseeable educational, privacy, security and misuse risks.

7.3 Privacy and security

Access to AI inputs, Outputs, logs and related Personal Data will be limited to authorised purposes and subject to reasonable technical and organisational safeguards.

7.4 Transparency

ScoreVedaa CAT will identify material AI-assisted functions where reasonably appropriate and explain their principal purposes and limitations. The Company is not required to disclose source code, proprietary system prompts, model weights, security controls, trade secrets, confidential provider information, or information whose disclosure would enable abuse or circumvention.

7.5 Human responsibility

AI does not remove the responsibility of Users to verify important information, Faculty Users to exercise educational judgment, Institutes to review material insights, or Company personnel to respond appropriately to verified issues.

7.6 Equality, inclusivity and non-discrimination

AI Features must not intentionally be designed or used to unlawfully discriminate against Users. The Company may use testing, monitoring, complaints and review processes to identify material bias or inappropriate proxy use.

7.7 Accountability

Material AI Features may have designated internal owners, provider records, purpose documentation, risk records and escalation processes proportionate to their risk.

7.8 Contestability

Where an Output or automated signal creates a material adverse consequence, the User will be provided access to a reasonable review or grievance mechanism where practicable and legally appropriate, except where immediate action or limited disclosure is reasonably necessary for security, safety, evidence preservation or compliance with Applicable Law.

7.9 Proportionality

Governance requirements may differ according to the purpose of the AI Feature, the sensitivity of the information, the likelihood and severity of harm, whether an Output is advisory or consequential, the degree of human involvement, the User category, the provider relationship, and Applicable Law.

8. AI Transparency, Labels and Provenance

ScoreVedaa CAT may identify AI-assisted functionality through interface labels, help text, tooltips, product descriptions, feature notices, this Policy, the Privacy Policy, or other reasonably accessible disclosures. A general AI disclosure does not mean that every individual Output was generated entirely by AI.

Where Applicable Law requires a Synthetic Content label, audible disclosure, metadata, unique identifier, provenance mechanism or related technical measure, ScoreVedaa CAT will implement the legally required measure to the extent applicable to the relevant feature and technically required under that law. A User must not knowingly remove, suppress, obscure, alter, defeat, or misrepresent a legally required Synthetic Content label, disclosure, identifier or provenance mechanism.

ScoreVedaa CAT may refuse to generate, publish, transmit or facilitate Synthetic Content that violates Applicable Law, deceptively impersonates a natural person, falsely portrays a real-world event, creates a false document or electronic record, violates another person’s privacy, contains unlawful sexual or intimate imagery, exploits or sexualises a minor, facilitates unlawful weapons or explosive activity, or otherwise creates an unacceptable legal or safety risk. Not every textual AI Output is required to contain an individual AI label unless Applicable Law or the relevant feature design requires one.

9. AI Features Used by ScoreVedaa CAT

ScoreVedaa CAT may use AI or Automated Processing for generating or assembling Assessments, suggesting question difficulty, classifying questions by topic or concept, analysing Assessment responses, calculating performance indicators, classifying topics, estimating Difficulty Coverage, identifying error patterns, suggesting possible root causes, generating Personalised Recommendations, prioritising topics, creating preparation plans, producing explanations, responding through PRGNA, analysing User uploads, summarising performance, supporting search and navigation, detecting misuse, detecting suspicious Account activity, content moderation, customer support, security monitoring, quality assurance, and other disclosed educational or operational purposes. The Company may introduce, modify, replace or withdraw an AI Feature in accordance with the Terms and this Policy.

10. AI Inputs and Data Sources

An AI Output may be based on one or more of the following: User prompts, User uploads, Assessment answers, Assessment scores, time and navigation information, Performance Data, User goals, Topic Classification, Difficulty Coverage, historical CAT information, ScoreVedaa Content, human-authored instructions, retrieval sources, rule-based logic, statistical methods, third-party AI models, Account context, User preferences, technical information, Cookie-derived information processed in accordance with the Cookie Policy, and other information lawfully processed under the Privacy Policy.

AI Outputs may be influenced by errors or limitations in User-provided information, Source Content, historical information, question labels, answer keys, model training data, retrieval results, Assessment design, system instructions, provider systems, and product configuration. The Company may not be able to identify or explain every technical factor influencing a particular Output. Where reasonably practicable, ScoreVedaa CAT may explain the primary educational indicators used for a material classification, recommendation or automated restriction.

11. Account-Level Personalisation

ScoreVedaa CAT may use available User information to customise or adapt the Services for the relevant User or Account. Account-Level Personalisation may include maintaining permitted conversational context, creating an Account-associated preparation profile, generating User-specific representations or embeddings, selecting relevant Assessments, ranking recommendations, adjusting explanation depth, adjusting question difficulty, recalculating action plans, identifying preparation gaps, adapting dashboards, and improving the relevance of Outputs for that User.

Account-Level Personalisation is distinct from training a Reusable Model for use across other Users. Where reasonably necessary to provide a requested or core feature, Account-Level Personalisation may occur without optional Training Consent. Depending on available functionality, a User may be able to modify goals, reset preferences, clear available conversational memory, reject a recommendation, disable an optional personalisation feature, or request assistance through support.

12. Human Review and Human Involvement

Not every AI Output is reviewed by a human before it is displayed. Human review may be used for reported errors, question validation, answer-key review, material Content issues, support, quality assurance, safety, security incidents, fraud investigations, copyright complaints, privacy complaints, grievances, material Account restrictions, optional model development covered by valid Training Consent, and product testing.

Human reviewers may include authorised Company personnel, Faculty or subject-matter reviewers, contractors, Data Processors, and professional advisers. Access will be subject to appropriate role restrictions, confidentiality obligations, data minimisation, security controls, retention limits, and purpose restrictions. Human review does not guarantee that an Output is correct.

Where a material restriction is based substantially on an automated security or integrity signal, ScoreVedaa CAT may conduct reasonable procedural or human review before permanent action, except where immediate restriction is reasonably necessary for security, safety, legal compliance or evidence preservation.

C. EDUCATIONAL AI FEATURES

13. Automated Performance Analysis

ScoreVedaa CAT may analyse questions presented, answers selected, correct, incorrect and unanswered responses, attempt rate, accuracy, time spent, questions skipped, questions revisited, navigation sequence, question difficulty, sectional performance, topic performance, Assessment history, performance variance, repeated error patterns, repeated topic avoidance, use of recommended activities, goal progress, and other disclosed educational signals.

The analysis may be used to generate reports, identify preparation patterns, classify topics, recommend activities, suggest possible causes of performance gaps, select Assessment difficulty, update goals, monitor progress, present readiness indicators, and improve the relevance of the User experience. Automated analysis may be affected by limited evidence, incorrect answers, guessing, interrupted sessions, Account sharing, unrepresentative Assessments, incorrect User information, question errors, technical problems, and changes in preparation behaviour. Automated analysis must not be treated as a complete evaluation of a User’s intelligence, capability, character or future potential.

14. Topic Classification

ScoreVedaa CAT may classify a topic using categories such as Strong, Inconsistent, Weak, Provisional, Unclassified, or another explained category. A classification may consider accuracy, attempt rate, Difficulty Coverage, evidence volume, evidence quality, recency, performance variance, question mix, time usage, Assessment integrity, and other disclosed indicators.

A Topic Classification is an internal preparation indicator, may be automated or AI-assisted, is not permanent, may change after new evidence, is not an academic credential, is not a diagnosis, does not determine official CAT performance, and must not be used as the sole basis for punishment or exclusion. Where evidence is insufficient, ScoreVedaa CAT may mark the classification as Provisional, withhold the classification, recommend further Assessments, use a lower validated level, or state that no reliable conclusion is available.

15. Difficulty Coverage

ScoreVedaa CAT may estimate the highest difficulty level or range at which a User has demonstrated reliable performance. Difficulty Coverage may consider accuracy thresholds, number of attempts, individual question difficulty, difficulty buckets, evidence quality, recency, consistency, question quality, and evidence sufficiency.

Difficulty labels are internal estimates. They may differ from labels assigned by teachers, coaching Institutes, publishers, other preparation platforms, or official examination authorities. Difficulty Coverage does not guarantee that the User will perform at the same level in the official CAT. Where evidence is limited, ScoreVedaa CAT may display a Provisional Output, recommend additional practice, use a lower validated level, or decline to display a result.

16. Personalised Recommendations and Action Plans

ScoreVedaa CAT may recommend actions such as Repair, Solidify, Stabilise, Maintain, Advance, Monitor, Deprioritise, Review a concept, Attempt a timed drill, Complete a foundation test, Complete a maintenance test, Attempt a challenge test, Reassess a topic, or Modify a preparation goal.

Recommendations may consider historical topic importance, Topic Classification, Difficulty Coverage, active goals, topic dependencies, days remaining, Assessment history, root-cause indicators, attempt behaviour, evidence sufficiency, and available study time. A recommendation is advisory. It may not account for health, stress, personal commitments, college or employment obligations, teacher guidance, disability or accessibility requirements, language barriers, financial constraints, or other circumstances not available to ScoreVedaa CAT.

Where supported, a User may accept, reject, skip, delay, replace, reattempt, change a goal, request review, or provide feedback. Rejecting a recommendation is not misconduct.

17. AI-Generated Assessments, Questions and Explanations

ScoreVedaa CAT may use AI to generate questions, adapt question wording, select questions, assemble Assessments, suggest answer options, generate explanations, assign topics, estimate difficulty, generate instructions, and produce draft solutions.

AI-generated or AI-assisted questions may contain incorrect facts, invalid assumptions, ambiguous wording, incorrect answer options, more than one defensible answer, no defensible answer, calculation errors, incorrect solutions, duplicate ideas, poor difficulty calibration, copyright concerns, or material unsuitable for the intended Assessment. AI-generated Assessments are not official CAT papers.

Where a material issue is identified, ScoreVedaa CAT may review, correct, replace, withdraw, disable, re-score, reclassify, or add clarification to the affected question or Assessment. A User reporting a suspected error should provide, where available, the Assessment name, the question identifier, the suspected issue, supporting reasoning, a screenshot, and any relevant source. A correction does not automatically entitle a User to monetary compensation except where required under the Terms, Refund Policy or Applicable Law.

18. PRGNA Learning and Doubt Assistance

PRGNA may provide concept explanations, worked examples, step-by-step solutions, question interpretation, reading-comprehension assistance, logical-reasoning guidance, quantitative methods, option-elimination guidance, study suggestions, responses concerning uploaded questions, and related educational support.

PRGNA may misinterpret a prompt, miss relevant context, use an unsuitable method, produce incorrect reasoning, produce an incorrect answer, invent unsupported information, overstate confidence, produce inconsistent responses, fail to recognise ambiguity, or conflict with a trusted solution. A User should independently verify a PRGNA response where the question is ambiguous, the input is incomplete, the answer conflicts with a trusted solution, the Output appears implausible, the issue materially affects preparation, or the User intends to rely on it in a consequential context.

PRGNA must not be used to obtain prohibited assistance during a live examination or another assessment where external assistance is forbidden. Private PRGNA conversations are not automatically shared with an Institute.

19. User Upload Analysis

A User may be permitted to upload questions, screenshots, notes, images, PDFs, study material, solutions, documents, and other authorised files for AI-supported explanation or analysis. Uploads may be processed to extract text, interpret a question, generate an explanation, identify a topic, suggest a method, provide support, detect prohibited Content, detect malware, investigate an error, maintain permitted context, or comply with law.

A User must not upload material the User has no right to use, confidential Institute material without authority, leaked examination questions, stolen question banks, Personal Data of another person without lawful authority, passwords, payment credentials, Government identifiers unless specifically requested, malware, illegal Content, material intended to manipulate or attack an AI System, or material prohibited by the Terms.

ScoreVedaa CAT does not guarantee that uploaded material will be interpreted correctly. Accuracy may be reduced by poor image quality, incomplete pages, handwriting, formatting, incorrect orientation, missing context, image compression, or unsupported file types. Users should retain independent copies of important uploads.

20. Provisional Outputs and Evidence Sufficiency

An AI Output may be marked Provisional, Low confidence, Insufficient evidence, Unavailable, Subject to review, or similar language. An Output may be provisional where too few questions were attempted, data is outdated, evidence is concentrated at one difficulty, performance is highly inconsistent, a technical issue affected the attempt, a question was invalid, Account sharing is suspected, the sample is unrepresentative, or relevant information is missing.

A Provisional Output must not be treated as a final judgment. Where reasonably practicable, ScoreVedaa CAT may explain why evidence is insufficient, what information is missing, which additional Assessment may assist, and whether the Output is likely to change. Absence of evidence is not proof of strength or weakness.

21. Confidence, Readiness and Similar Indicators

ScoreVedaa CAT may display indicators described as Confidence, Readiness, Mastery, Coverage, Preparation strength, Goal progress, or similar terms. Such indicators may consider topic importance, Classification, Difficulty Coverage, evidence quality, recency, Assessment history, goal scope, and behavioural indicators.

These indicators do not represent emotional confidence, psychological status, intelligence, mental health, an official CAT score, an official percentile, admission probability, selection certainty, or guaranteed future performance. A maximum internal value does not guarantee complete knowledge or examination success. Indicators may change after additional Assessments, question corrections, updated logic, goal changes, evidence review, provider changes, or product updates.

22. Historical Information and Prediction Limitations

ScoreVedaa CAT may use historical CAT information to identify topic trends, estimate importance, analyse frequency, compare difficulty, inform priorities, support preparation planning, and generate contextual explanations.

Historical patterns do not guarantee future patterns. Past topic frequency, question type, difficulty, weightage, sectional distribution, score-to-percentile relationships, admission cut-offs, and examination structure may not repeat. ScoreVedaa CAT does not claim access to confidential future CAT papers, leaked questions, unreleased official examination data, or non-public admission decisions. A recommendation based on historical information must not be interpreted as a certain prediction.

23. Beta and Experimental AI Features

A beta, pilot or experimental AI Feature may be incomplete, have limited testing, produce unstable Outputs, contain material errors, have lower availability, change frequently, or be withdrawn. A beta or experimental designation may be displayed where reasonably practicable. Users must not rely on a beta feature for official CAT information, admission certainty, high-stakes preparation decisions without verification, legal, medical or financial advice, or a purpose outside the disclosed testing scope.

ScoreVedaa CAT may restrict beta access to selected Users. Feedback concerning a beta feature may be used for evaluation, error correction, safety, reliability, feature design, and operational improvement. Private User Content will not be used for identifiable Reusable Model development merely because it was submitted through a beta feature.

D. LIMITATIONS, RELIANCE AND USER RESPONSIBILITIES

24. General AI Limitations

AI Systems have inherent limitations. AI Outputs may be affected by incomplete data, poor-quality uploads, ambiguous questions, incorrect labels, model bias, hallucination, statistical uncertainty, context limitations, language limitations, prompt interpretation, retrieval errors, provider failure, technical failure, changes in examination patterns, source errors, and human configuration errors.

An AI System may produce a confident but incorrect response, fail to recognise insufficient information, produce different responses to similar prompts, misunderstand the User’s circumstances, omit relevant assumptions, reproduce common or similar phrasing, or fail to recognise that Content may be protected by third-party rights. No AI Output should be treated as infallible.

25. No Guarantee of Accuracy or Availability

To the maximum extent permitted by Applicable Law, ScoreVedaa CAT does not guarantee that an AI Output will always be correct, complete, current, original, unbiased, error-free, consistent, suitable for a particular purpose, equivalent to teacher guidance, human-reviewed, or available without interruption. ScoreVedaa CAT will use reasonable processes proportionate to the relevant feature to improve reliability, monitor material issues, respond to reports, correct verified problems, and manage provider or system failures. Nothing in this Section excludes responsibility for fraud, wilful misconduct, mandatory consumer remedies, mandatory data-protection obligations, or liability that cannot lawfully be excluded.

26. No Guarantee of CAT Results or Admission

AI Outputs do not guarantee a minimum CAT score, a raw score, a scaled score, a percentile, a sectional cut-off, an interview call, admission, a scholarship, selection, offer conversion, or improvement within a stated period. Examination and admission outcomes depend on factors outside ScoreVedaa CAT’s control, including examination difficulty, official scaling, competition, preparation consistency, health, stress, time management, academic records, work experience, interviews, institution-specific rules, and actual examination-day performance. Any estimate, range, prediction, readiness indicator or benchmark is illustrative and non-binding.

27. User Responsibility

Users remain responsible for reviewing AI Outputs, verifying important information, following official CAT notifications, deciding which recommendations to follow, determining study time, seeking teacher support where appropriate, protecting Account credentials, avoiding prohibited submissions, respecting intellectual-property rights, avoiding cheating, providing accurate information, and reporting serious errors.

Users should not rely exclusively on AI for every preparation decision, ignore official examination information, continue a clearly unsuitable plan, submit unnecessary sensitive information, treat a classification as permanent, treat an Output as an official CAT statement, represent AI Output as independently verified, use AI to impersonate another person, or use an Output outside its stated educational context without appropriate review. A User should stop relying on and report an Output that appears clearly incorrect, harmful, inappropriate, unlawful, discriminatory, manipulated, based on misunderstood information, or inconsistent with a trusted source.

28. Institute and Faculty Use

Institutes and Faculty Users may use authorised AI Outputs to support educational review, identify topics for discussion, recommend practice, monitor assigned activity, assist mentoring, support authorised Student programmes, and review aggregate programme trends.

Institutes and Faculty Users must not use an AI Output as the sole basis for punishment, expulsion, public shaming, admission denial, employment action, financial discrimination, medical or psychological judgment, labelling a Student as incapable, publishing a ranking without authority, or another unrelated High-Impact Decision. Institute personnel must independently review material AI insights before relying on them.

An Institute must not represent that ScoreVedaa CAT guarantees results, AI Outputs are official CAT assessments, a classification is permanent, a recommendation is compulsory unless independently assigned, every Output has been human-reviewed, or ScoreVedaa CAT automatically shares all Student data. An Institute remains responsible for its independent decisions and Processing outside ScoreVedaa CAT.

29. Student and Institute Data Visibility

ScoreVedaa CAT does not automatically provide an Institute with unrestricted access to private PRGNA conversations, private prompts, personal uploads, direct support communications, private study notes, payment credentials, passwords, unrelated personal goals, cookie choices, general browsing activity, or activity outside the authorised programme.

An Institute or Faculty User may view selected AI-related information where the Adult Student expressly shares it, an authorised share function is used, the Student accepts an Institute invitation after receiving notice, the information relates to an authorised programme and access has been disclosed, a lawful institutional arrangement permits access, or disclosure is required by law.

A Student displaying an AI Output to a Faculty User does not automatically grant permanent access, a continuing data feed, access to future Outputs, a right to copy unrelated information, or ownership of the information. Institute personnel must handle AI-related Student information in accordance with the Privacy Policy, the Institute Agreement, role-based access restrictions, confidentiality obligations, Applicable Law, and authorised educational purposes.

30. High-Impact Decisions and Automated Restrictions

ScoreVedaa CAT’s educational AI Features are not intended to independently make High-Impact Decisions. ScoreVedaa CAT will not provide an Institute with an AI Feature intended to serve as the sole basis for admission, employment, lending, insurance, Government benefits, medical treatment, psychological diagnosis, or another materially consequential determination unrelated to CAT preparation.

Automated security, integrity or misuse signals may temporarily result in additional verification, prompt blocking, upload blocking, session termination, payment review, attempt review, or temporary Account restriction. Where a permanent material restriction is based substantially on an automated signal, the User will be provided access to a reasonable review or grievance mechanism where practicable and legally appropriate, except where immediate action or limited disclosure is reasonably necessary for security, safety, evidence preservation or compliance with Applicable Law. Immediate action may be taken where reasonably necessary to protect security, prevent continuing harm, preserve evidence, comply with law, protect intellectual property, or prevent serious misuse.

E. PERSONAL DATA, PERSONALISATION AND MODEL IMPROVEMENT

31. Personal Data Processing Through AI

AI Processing involving Personal Data is governed by the Privacy Policy and Applicable Law. ScoreVedaa CAT may process Personal Data through AI to provide a requested explanation, generate a requested Output, analyse performance, personalise recommendations, generate action plans, maintain permitted context, operate Assessments, secure the Platform, detect abuse, provide support, investigate errors, meet legal obligations, and perform other disclosed purposes.

ScoreVedaa CAT will apply, as applicable: appropriate notice, purpose limitation, data minimisation, Consent where required, security safeguards, retention controls, Data Processor controls, User-rights processes, and grievance mechanisms. Users should not submit unnecessary Government identifiers, payment credentials, health information, passwords, private third-party information, confidential commercial information, legally privileged information, or other sensitive information unrelated to the requested Service. ScoreVedaa CAT may remove, mask, restrict, quarantine or decline to process information that appears unnecessary, unsafe or prohibited.

32. AI Inference Is Distinct From Model Training

ScoreVedaa CAT may transmit or process relevant prompts, conversations, uploads, Assessment responses, Performance Data, instructions, Account context, goals, and technical information to generate an immediate AI Output or provide a requested AI Feature. This may include Processing by authorised Company systems and Service Providers.

Processing information to respond to a prompt, generate an explanation, analyse an Assessment, produce a recommendation, maintain permitted context, moderate an input, detect abuse, or troubleshoot a feature is distinct from using that information to train a Reusable Model. A User’s decision to use PRGNA or another AI Feature does not automatically authorise Reusable Model training.

33. Irreversibly Anonymised and Aggregated Improvement

The Company may transform Personal Data into information that is irreversibly anonymised or aggregated so that it no longer identifies or can reasonably be linked to an individual. Irreversibly anonymised or aggregated information may be used for product analytics, reliability measurement, capacity planning, safety evaluation, error-rate analysis, research, benchmarking, Assessment improvement, recommendation improvement, internal model training, internal model evaluation, and internal product development.

Removing a name, email address, phone number or Account identifier does not by itself constitute irreversible anonymisation. Measures may include removal of direct identifiers, generalisation of indirect identifiers, free-text filtering, aggregation, thresholding, tokenisation, separation of lookup keys, access controls, re-identification risk assessment, and restrictions on re-identification. The Company will not intentionally re-identify information represented as irreversibly anonymised. An individual contribution may not be removable where it has been genuinely and irreversibly incorporated into an aggregate, statistical results, evaluation metrics, or model parameters that cannot reasonably identify or be linked to the User.

34. Optional Internal Reusable Model Development

The Company may invite a User to provide separate optional Training Consent for internal Reusable Model development. Where valid Training Consent is provided, the Company may use the specified categories of Private User Content or identifiable Performance Data to train, fine-tune, test, evaluate, validate, calibrate, benchmark and improve Company-controlled ScoreVedaa or NeuraMach AI Systems. Purposes may include improving CAT question generation, explanation quality, topic classification, difficulty estimation, error detection, recommendation quality, personalisation systems, retrieval systems, safety systems, Assessment assembly, performance analysis, and other CAT educational capabilities described in the Model Improvement Notice.

Before Training Consent is obtained, the Model Improvement Notice must identify, as applicable: the proposed data categories, whether prompts are included, whether PRGNA conversations are included, whether uploads are included, whether Assessment responses are included, whether Performance Data is included, whether historical information is included, the model-development purposes, whether human review may occur, whether approved Data Processors may assist, the retention approach, the withdrawal mechanism, and material consequences of refusal or withdrawal.

Training Consent is separate from Terms acceptance, is separate from Privacy Policy acknowledgement, requires a clear affirmative action, must not be pre-selected, is optional, does not affect access to core Services, is recorded, does not authorise provider-side training, and may be withdrawn. Where the User does not provide Training Consent, identifiable Private User Content and Performance Data will not be selected for optional internal Reusable Model development. Identifiable CAT information will not be used for cross-product Reusable Model development unless the cross-product scope is clearly disclosed, the relevant data and purposes are identified, any required separate affirmative Consent is obtained, and appropriate safeguards are maintained.

35. Withdrawal of Training Consent

A User may review or withdraw Training Consent through: Settings → Privacy & AI → AI Model Improvement where that functionality is available. The relevant setting may be labelled substantially: “Use my identifiable data to improve ScoreVedaa’s AI models.” Training Consent may also be withdrawn through web Account settings, mobile-app settings, an applicable consent-management interface, privacy@neuramach.ai, or another communicated method.

Following withdrawal, the Company will within a reasonable period: stop selecting new identifiable information for the withdrawn purpose, update the User’s consent status, update dataset-eligibility records, remove or suppress identifiable information from active training or evaluation datasets where reasonably practicable and legally required, instruct relevant Data Processors to stop the withdrawn optional Processing, and retain only information required for legal, audit, security, fraud or dispute purposes.

Withdrawal applies prospectively. Withdrawal does not affect Processing lawfully completed before withdrawal. Individual removal may not be possible where information has already been irreversibly anonymised, included in a completed aggregate, incorporated into completed statistical results, or incorporated into model parameters in a manner that cannot reasonably identify or be linked to the User. Withdrawal of Training Consent does not automatically disable Account access, PRGNA, Assessment services, AI inference, performance analysis, core Account-Level Personalisation, or other Processing reasonably necessary to provide the Services.

36. Limited Quality Assurance, Safety and Incident Review

Regardless of Training Consent, the Company may conduct limited and purpose-specific review of a particular prompt, conversation, AI Output, Assessment response, performance record, upload, security event, or related context where reasonably necessary to reproduce a reported error, respond to support, investigate an AI Incident, detect fraud or abuse, verify whether a feature operated as represented, review a question or answer key, investigate a classification or recommendation, respond to a legal complaint, test a specific correction, enforce the Terms, protect security, or protect the reliability and safety of the Services.

This Processing may be used to correct software defects, correct questions or answer keys, correct retrieval sources, correct rules or calculations, adjust operational prompts, adjust workflows, improve rule-based safeguards, resolve the incident, or produce anonymised operational metrics. This limited review does not independently authorise general identifiable-data Reusable Model training.

37. Third-Party AI and Cloud Providers

ScoreVedaa CAT may use authorised third-party providers for large language models, AI inference, model hosting, cloud hosting, retrieval infrastructure, text extraction, content moderation, security, logging, error monitoring, customer support, and related infrastructure. Such providers may process relevant prompts, conversations, User Content, Performance Data, Account context, and technical information only as reasonably necessary for an authorised Service or purpose.

The Company will use only Paid, Enterprise, Business, API, or other contractually controlled configurations under which binding written terms, applicable data-processing terms or enforceable service settings prohibit the provider from using Private User Content or identifiable Performance Data for the provider’s own general-purpose model training, independent product improvement, independent model fine-tuning, general provider benchmarking, or unrelated purposes, unless separate provider-side use has been disclosed and affirmatively consented to under Section 38.

The Company will not route live Private User Content or identifiable Performance Data through free consumer AI accounts, unpaid configurations permitting provider training, personal chatbot accounts, public playgrounds, unapproved experimental endpoints, or provider configurations whose data-use terms have not been reviewed.

Before routing live information through a material provider, the Company will take reasonable steps to review contractual terms, data-processing terms, retention settings, training settings, human-review conditions, security controls, sub-processors, international Processing, incident handling, and service configuration. If the Company becomes aware that required protections no longer apply, it will within a reasonable period stop routing new live identifiable information through the affected configuration, restore a compliant configuration, move Processing to another compliant provider, or suspend the affected feature where reasonably necessary. A provider may process limited information for security, abuse detection, moderation, troubleshooting, reliability, legal compliance, or service operation where contractually permitted, reasonably necessary and lawful.

38. Provider-Side Model Training

Internal Training Consent does not authorise a third-party provider to use Personal Data for the provider’s own model training, independent fine-tuning, independent benchmarking, general provider product development, or improvement of services not supplied exclusively for the Company.

Where ScoreVedaa CAT proposes provider-side model training or independent provider product improvement in the future, the Company must first provide a separate Provider Training Notice identifying, as applicable: the provider or provider category, the relevant data categories, the provider’s purpose, whether the provider acts independently, retention, human review, international Processing, withdrawal methods, and other material information required by Applicable Law. Separate affirmative Consent must be obtained before enabling provider-side training or independent provider product improvement using Private User Content or identifiable Performance Data.

Provider-side Consent must be separate from internal Training Consent, must not be inferred from Terms acceptance, must not be inferred from AI Feature use, must not be inferred from continued use, must not be represented by a pre-enabled control, must be recorded, must be withdrawable, and must identify the provider use with sufficient clarity. Refusal to provide provider-side Consent will not prevent access to ordinary ScoreVedaa CAT Services where a compliant no-provider-training configuration is reasonably available.

39. Cookies, Technical Identifiers and AI

Use of Cookies and Similar Technologies is governed by the Cookie Policy. Cookie Consent does not authorise internal identifiable-data Reusable Model training, provider-side model training, cross-product model development, human review for general model development, or unrelated profiling.

Identifiable Cookie-derived information will not be selected for optional internal model development unless the Model Improvement Notice includes the relevant data category, the specified purpose covers the Processing, and valid Training Consent has been provided. Cookie-derived Personal Data will not be used for provider-side training unless a Provider Training Notice has been provided, separate affirmative Consent has been obtained, and the Processing complies with the Privacy Policy and Applicable Law. Irreversibly anonymised or aggregated technical information may be used in accordance with the Privacy Policy.

F. SAFETY, ACCEPTABLE USE AND INTELLECTUAL PROPERTY

40. Prohibited AI Uses

Users must not use an AI Feature to:

  • Cheat in an active examination;
  • Obtain, process or distribute leaked examination material;
  • Impersonate another person;
  • Create a deceptive deepfake or false electronic record;
  • Falsely portray a natural person or real-world event;
  • Create fraudulent academic records;
  • Harass, threaten or abuse;
  • Generate unlawful discriminatory Content;
  • Create non-consensual intimate imagery;
  • Create sexual or exploitative material involving minors;
  • Facilitate self-harm or violence;
  • Facilitate unlawful weapons, explosives or ammunition activity;
  • Commit fraud;
  • Generate or distribute malware;
  • Circumvent security;
  • Conduct prompt injection or model manipulation;
  • Extract source code;
  • Extract confidential system prompts;
  • Attempt model theft;
  • Scrape protected Content;
  • Create a competing question bank through systematic extraction;
  • Violate copyright;
  • Submit another person’s Personal Data without lawful authority;
  • Manipulate scores or Performance Data;
  • Circumvent usage limits;
  • Conduct unauthorised automated testing;
  • Misrepresent an AI Output as official CAT guidance;
  • Remove a legally required Synthetic Content label;
  • Make an unlawful High-Impact Decision; or
  • Violate Applicable Law.

The Company may block or restrict a prompt, an upload, an Output, a conversation, a feature, a session, an Account, or an integration where misuse is reasonably suspected.

41. Safety Controls and Content Moderation

ScoreVedaa CAT may use automated and human controls to detect harmful prompts, block prohibited Content, restrict unsafe Outputs, detect malware, prevent cheating, identify possible leaked Content, detect attempts to bypass safeguards, detect prompt injection, protect individuals from unlawful Synthetic Content, protect intellectual property, preserve legally required labels, and maintain Platform security.

Safety systems may refuse a request, limit a response, provide a warning, request clarification, remove or quarantine an upload, apply a label, escalate for review, restrict access, preserve relevant records, or notify an appropriate recipient where legally required. Safety systems may make mistakes. A refusal does not necessarily establish misconduct. A User may report an incorrect refusal through support or the grievance process. The Company may preserve information concerning serious abuse where reasonably necessary for investigation, security, legal compliance, evidence preservation, dispute resolution, protection of Users, or enforcement of the Terms.

42. Academic Integrity

Users must comply with the rules applicable to any examination, Assessment, Institute programme or educational activity. AI Features must not be used during a live CAT examination, during another examination prohibiting external assistance, to impersonate a candidate, to manipulate Assessment records, to obtain leaked questions, to produce fraudulent submissions, or to misrepresent independent work. ScoreVedaa CAT may use available technical and behavioural indicators to investigate suspected misuse. An automated indicator alone may not be conclusive proof of cheating. The Company may consider technical logs, timing, Account access, attempt history, User explanations, available evidence, and human review before permanent action where reasonably practicable.

43. Confidentiality and Sensitive Information

Users should not submit to an AI Feature passwords, UPI PINs, CVVs, full card information, bank passwords, verification codes, confidential business plans, unauthorised Institute material, third-party private information, health records, Government identifiers, legally privileged material, trade secrets, or information unrelated to CAT preparation.

ScoreVedaa CAT may use automated or manual controls to detect, mask, block, quarantine, restrict, or remove prohibited or unnecessary information. The Company does not guarantee that every sensitive submission will be detected before Processing. A User who accidentally submits sensitive information should promptly contact privacy@neuramach.ai or support@neuramach.ai. Institute personnel must not submit Student Personal Data into an AI Feature beyond what is authorised and reasonably necessary.

44. Intellectual Property and AI Outputs

ScoreVedaa Content, software, workflows, interfaces, system prompts, models, test structures, reports and related intellectual property remain owned by or licensed to the Company. Use of an AI Feature does not transfer ownership of the underlying model, source code, confidential prompts, algorithms, question banks, proprietary reports, Platform logic, evaluation methods, or training systems. A User retains rights lawfully held in User Content, subject to the licences and restrictions stated in the Terms.

AI Outputs may contain similarities to existing material, reproduce common phrases, be unoriginal, be subject to third-party rights, be unsuitable for commercial use, lack independent copyright protection, or contain material that requires verification or attribution. ScoreVedaa CAT does not guarantee that every AI Output is unique, non-infringing, registrable, commercially exploitable, or exclusively owned by the User. Users must not use AI Features to infringe copyright, reproduce paid coaching material, generate derivative question banks from protected Content, remove attribution, misrepresent authorship, circumvent technological protections, resell ScoreVedaa Content, or use Outputs in violation of the Terms.

45. Bias, Fairness and Sensitive Inferences

AI Outputs may reflect limitations or bias arising from historical information, provider training data, question selection, language, topic labels, User behaviour, sample size, model design, human assumptions, retrieval sources, or product configuration. The Company will not intentionally design AI Features to unlawfully discriminate on grounds protected by Applicable Law. Educational-performance analysis should not infer or determine caste, religion, race, ethnicity, political opinion, sexual orientation, disability, medical status, economic status, or another unrelated sensitive characteristic unless Processing is legally permitted, genuinely necessary for a disclosed purpose and subject to appropriate safeguards. The Company may conduct proportionate review or testing to identify material performance disparities, systematic error, language disadvantage, inappropriate proxies, repeated harmful Outputs, unfair exclusion, and other fairness concerns. A User may report suspected bias through the grievance process.

G. AI GOVERNANCE, SECURITY AND INCIDENTS

46. Monitoring and Operational Review

ScoreVedaa CAT may monitor AI Features for reliability, security, abuse, quality, error rates, safety incidents, provider performance, data leakage, prompt injection, copyright complaints, Synthetic-content misuse, privacy concerns, and compliance. Monitoring may involve automated logs, limited prompt and response sampling, error reports, security alerts, User feedback, provider reports, human review, test datasets, and internal testing. Access to Private User Content for review will be limited to authorised personnel with a legitimate purpose. Monitoring does not independently authorise general identifiable-data Reusable Model training. Monitoring records will be retained in accordance with the Privacy Policy, Applicable Law, security requirements, incident needs, and applicable retention schedules.

47. AI Risk Assessment and Governance

Before launching or materially changing a Material AI Feature, the Company will conduct a proportionate review having regard to the nature, purpose and reasonably foreseeable risks of that feature.

Depending on the risk, safeguards may include additional testing, restricted access, human review, enhanced notices, limited data use, provider controls, staged release, monitoring, approval by designated personnel, Output restrictions, usage limits, or suspension pending remediation. The governance level may differ between a low-risk spelling or navigation assistant, a performance recommendation, a question-generation system, a fraud-detection system, a Synthetic Content generator, and a feature capable of materially restricting an Account. A Material AI Feature will not be deployed or continued where identified risks cannot be reduced to a level considered acceptable under Applicable Law and the intended educational purpose.

48. AI and Provider Records

The Company will maintain proportionate internal records for Material AI Features, having regard to their nature, purpose and risk. Such records may include feature ownership, intended purpose, model or provider information, data categories, data flows, training status, consent requirements, provider configurations, risk reviews, test results, known limitations, incident records, change records, complaint records, and approval records.

Governance records are internal controls and not themselves the legal basis for Processing. However, material discrepancies identified in governance records will be assessed and corrected within a reasonable period. The Company may protect internal governance records from public disclosure where they contain trade secrets, security information, provider-confidential information, privileged legal advice, Personal Data, or information whose disclosure would enable misuse.

49. Security of AI Systems

The Company will use reasonable technical and organisational measures appropriate to relevant risks. Measures may include encryption in transit, encryption at rest where appropriate, access controls, role-based permissions, secret and key management, logging, monitoring, vendor assessment, network protection, secure-development practices, vulnerability management, input validation, Output filtering, rate limiting, backup controls, incident response, dataset controls, environment separation, provider-configuration review, and staff confidentiality obligations.

Users must protect Account credentials, avoid sharing API keys or tokens, avoid uploading malware, avoid bypassing safeguards, report suspected vulnerabilities responsibly, and avoid publicly disclosing an unremediated vulnerability in a manner that creates further risk. No system can be guaranteed completely secure.

50. AI Incidents

An AI Incident may include materially harmful Output, repeated materially incorrect answers, exposure of Private User Content, unauthorised model-training use, security compromise, model manipulation, prompt injection, copyright reproduction, discriminatory Output, Synthetic-content misuse, provider breach, data leakage, failure of safety controls, or a serious misleading recommendation.

Where an AI Incident occurs, the Company may contain the issue, disable or restrict the feature, investigate, preserve evidence, notify providers, correct Outputs, re-score Assessments, reset access, change providers, update safeguards, correct governance records, and implement remediation. The Company will provide notifications to affected Users, authorities and other legally required recipients within the periods required by Applicable Law. Where an AI Incident constitutes or involves a Personal Data breach, the Privacy Policy’s breach-response framework applies. Users should report a suspected AI Incident promptly through the contact information in Section 58.

51. AI Feature Changes and Versioning

AI Features may change because of model updates, provider changes, prompt changes, retrieval changes, safety updates, new data sources, revised scoring logic, updated question banks, legal requirements, security changes, performance improvements, or product redesign. As a result, Outputs may change, previous recommendations may differ, difficulty labels may be recalculated, classifications may change, explanations may be updated, safety behaviour may change, and features may be withdrawn.

Where a material change affects Personal Data use, Reusable Model training, provider-side training, Institute visibility, High-Impact automated use, Synthetic Content, a material provider relationship, User rights, or the nature of AI Processing, the Company will provide updated notice and obtain Consent where required by Applicable Law. The Company may retain prior logic, policy versions, prompt versions, provider records, evaluation results, and model-version information for audit, dispute resolution, safety and compliance purposes.

52. Third-Party Content and External AI Tools

ScoreVedaa CAT may display or link to Content produced by publishers, Institutes, teachers, Content providers, AI providers, external websites, and other third parties. Third-party Content may be subject to separate terms, separate privacy policies, separate accuracy limitations, intellectual-property rights, provider retention, provider training practices, and independent moderation. ScoreVedaa CAT is not responsible for an external AI tool used independently by a User.

Users should not transfer Private User Content from ScoreVedaa CAT into an external AI system without considering privacy, confidentiality, intellectual property, provider retention, model training, Institute restrictions, Applicable Law, and whether the User has authority to make the disclosure. A link to an external AI tool does not constitute endorsement unless expressly stated.

H. REPORTING, REVIEW, ENFORCEMENT AND CONTACT

53. Reporting Errors and Harmful Outputs

Users may report incorrect answers, incorrect explanations, ambiguous questions, harmful Content, bias, inappropriate language, privacy concerns, copyright concerns, leaked Content, unlawful Synthetic Content, misclassification, incorrect recommendations, data exposure, security concerns, unauthorised model training, or other AI problems.

A report should include, where available, Account identifier, feature name, date and time, Assessment or conversation identifier, description of the issue, screenshot, supporting explanation, and requested resolution. Users should avoid including additional sensitive information unless necessary. The Company may investigate, request further information, correct an Output, remove Content, re-score, reclassify, restrict a feature, notify a provider, preserve evidence, decline an unsupported request, or provide another appropriate response. Submitting a report does not guarantee that the Company will agree with the User’s academic interpretation.

54. User Review and Contesting Outputs

Where supported, a User may contest or seek review of a Topic Classification, Difficulty Coverage, a recommendation, a readiness indicator, a flagged Assessment, a generated answer, a score affected by a question error, a material Account restriction, or another material AI Output. Review may involve an additional Assessment, examination of source data, verification of answer keys, technical-log review, human academic review, correction of User information, recalculation, provider investigation, or confirmation that the original Output remains unchanged.

The Company may decline repeated requests raising no new information, abusive requests, requests intended to obtain confidential model information, requests that compromise security, or requests outside the applicable review process. Where an Output has a material adverse consequence, the User will be provided access to a reasonable review or grievance mechanism where practicable and legally appropriate, except where immediate action or limited disclosure is reasonably necessary for security, safety, evidence preservation or compliance with Applicable Law.

55. Suspension and Restriction of AI Access

ScoreVedaa CAT may restrict AI Features where a User violates this Policy, misuse is reasonably suspected, security is threatened, a payment entitlement expires, a feature is unstable, a provider becomes unavailable, legal compliance requires restriction, copyright concerns arise, Synthetic-content misuse occurs, a serious incident is investigated, or continued use creates unreasonable risk.

A restriction may involve prompt blocking, upload blocking, Output blocking, reduced usage, conversation suspension, feature suspension, Account verification, temporary Account restriction, or permanent termination for serious or repeated misuse. Where appropriate, the Company may provide notice, a reason, an opportunity to respond, a review process, and a grievance channel. Immediate action may be taken for urgent security, legal, safety, privacy, intellectual-property, or evidence-preservation risks.

56. AI-Related Grievances

AI-related grievances may be submitted to:

Grievance Officer: Mr. Saurabh Sachdeo

Designation: Grievance Officer - Legal and Compliance

Company: NEURAMACH AI STUDIO PRIVATE LIMITED

Email: grievance@neuramach.ai

Phone: +91-8223815327

Address: 3rd Floor, Cabin No. 7, Quick Office, 301, 45 Baner Road, above Atithi Restaurant, Veerbhadra Nagar, Baner, Pune, Maharashtra 411045, India.

A grievance should include Name, registered contact information, Account identifier, AI Feature involved, description of the issue, relevant dates, supporting material, and requested resolution. For AI-related privacy or consumer grievances not governed by a shorter statutory period, the Company will acknowledge the grievance within forty-eight hours and resolve it within one month of receipt, subject to Applicable Law. Where Applicable Law requires a shorter period, the shorter period applies.

Where further information is reasonably required, the Company may request it. Where exceptional complexity or a legal dependency prevents resolution within the stated period, the Company may provide the current status, the reason for delay, further information required, and an expected completion date, without limiting a mandatory legal deadline. Where an AI-related complaint is governed by the intermediary grievance requirements under Applicable Law, the applicable twenty-four-hour acknowledgement, seven-day resolution or other shorter statutory deadline will apply.

Urgent issues involving Personal Data exposure, security compromise, harmful Synthetic Content, unauthorised model training, active examination misconduct, serious copyright infringement, or another continuing risk will be prioritised according to severity. Content-removal and intermediary complaints may be subject to the separate timelines stated in the Terms and Conditions.

If your grievance is not resolved satisfactorily through our process, or the response period lapses, you may lodge a complaint with the Data Protection Board of India in the manner it prescribes.

57. Changes to this Policy

The Company may update this Policy to reflect new AI Features, new providers, new data uses, legal changes, safety changes, model changes, Institute-feature changes, Synthetic-content requirements, security improvements, governance changes, or clarifications. The updated Policy will display a revised version number, an effective date, and a last-updated date.

For material changes involving new use of Private User Content, internal identifiable-data model training, provider-side model training, historical-data use, cross-product model development, new profiling purposes, material Institute visibility, High-Impact automated use, Synthetic Content, new material providers, new categories of Personal Data, or materially reduced User rights, the Company will provide prominent notice and obtain renewed Consent where required by Applicable Law.

A Policy update, continued use, silence or failure to opt out will not constitute Training Consent or provider-side model-training Consent where separate affirmative Consent is required. Previously collected Private User Content will not be used for a newly introduced identifiable-data model-training purpose merely because this Policy is updated. Non-material clarifications may take effect upon publication. Earlier versions may be retained for compliance, audit, evidence and dispute-resolution purposes.

58. Contact Information

For AI questions, reports, privacy requests or legal concerns, contact:

Legal Entity: NEURAMACH AI STUDIO PRIVATE LIMITED

Brand: NeuraMach.ai

Product: ScoreVedaa CAT

CIN: U62099PN2025PTC245340

Website: scorevedaa.in

Registered Office: Flat No. 201, Building 1, Wing 3, The Crown Greens, Plot 17, Infotech Park, Hinjawadi, Pune, Maharashtra 411057, India.

Principal Business and Correspondence Office: 3rd Floor, Cabin No. 7, Quick Office, 301, 45 Baner Road, above Atithi Restaurant, Veerbhadra Nagar, Baner, Pune, Maharashtra 411045, India.

AI and General Support: support@neuramach.ai

Privacy: privacy@neuramach.ai

Security: security@neuramach.ai

Grievance: grievance@neuramach.ai

Legal: legal@neuramach.ai

Telephone: +91-8223815327

Grievance Officer

Name: Mr. Saurabh Sachdeo

Designation: Grievance Officer - Legal and Compliance

Email: grievance@neuramach.ai

Phone: +91-8223815327