AI transparency
AI Usage Policy
This policy explains the separate jobs of Magic AI and PRGNA, the learning context they may use, the difference between reviewed bank content and a live generated reply, and the choices available to each learner.
In brief
Magic AI
Strategy and orchestration: it helps turn available evidence into a next action.PRGNA
Teaching and guided practice: it explains the selected gap and supports follow-up work.Not a guarantee
AI can be wrong and does not determine eligibility, results, jobs or selection.01Two AI roles, kept distinct
Magic AI helps with preparation strategy and next actions. It can help summarise available attempt evidence, identify a useful priority and suggest the next practice, revision or review action. It should not pretend that an unattempted area is a proven weakness.
PRGNA is the teaching and guided-practice layer. It explains the topic or error you choose, adapts an explanation within the available learning context and can hand a useful next action back to the preparation flow. Magic AI does not replace PRGNA’s teaching role, and PRGNA does not invent a full strategy from unsupported assumptions.
Back to top02When AI becomes available
Magic AI and PRGNA are currently gated for a selected course until that course has usable, reviewed learning content and the account has valid access. Creating an identity account, verifying an OTP or saving a course does not activate either AI feature or start a trial.
When AI access opens, ScoreVedaa checks that the signed-in account, selected IBPS course, course access and the separate AI Usage consent are active before processing a learning request. Magic AI and PRGNA use only the learner's selected IBPS course context.
Back to top03What the AI can use
Depending on the request and available features, the AI may use:
- the message you type and the topic, question or task you selected;
- your verified IBPS course, stage, section timing and navigation rules;
- completed attempt evidence such as accuracy, time, errors and review state;
- goals or preferences that you saved for preparation;
- recent conversation context needed for a coherent follow-up; and
- rights-cleared ScoreVedaa explanations or question-bank content that has passed review.
Magic AI should distinguish evidence from inference. When there is not enough evidence, it should ask for a diagnostic attempt or state the uncertainty instead of manufacturing a weakness, score or probability.
Back to top04What an AI answer means
A live Magic AI or PRGNA response is generated on demand and may not be reviewed by a person before you see it. It may be incomplete, outdated, unclear or wrong. Questions and solutions published by ScoreVedaa are different: they must pass authorship, rights and academic review before learners see them.
AI responses are preparation assistance, not official IBPS information, legal or financial advice, recruitment counselling, an eligibility decision or a human promise. Check examination dates, eligibility, syllabus, rules and results against official authority announcements. Report a suspect answer from the relevant answer review or support channel so it can be investigated.
Back to top05Planned PRGNA text language choices
When this capability is released, PRGNA will offer text explanations in up to 28 language choices. It will be specific to PRGNA text. It will not mean that the entire interface, a question paper, a solution bank, voice output, support response or legal document is available in every language.
Exam terms may remain exactly as used in the reviewed paper configuration. A bilingual or translated paper is released only after every required version is complete and reviewed; a machine-generated translation is not treated as publication-ready merely because PRGNA can converse in that language.
Back to top06AI data protection
Before production AI becomes available, NeuraMach must verify an approved cloud service in the Mumbai region, its processor terms and its regional evidence. Until that release gate is complete, Magic AI and PRGNA remain unavailable. An approved release must block cross-region inference profiles, model routes outside India and automatic fallback to another country; if the approved India-region path is unavailable, the feature must return an unavailable state instead of sending a request to another model or region.
Before a request, ScoreVedaa masks high-confidence identifiers in the message and supplied context, including email addresses, Indian mobile numbers and recognisable identity or payment identifiers. This reduces exposure but does not make arbitrary free text anonymous. Messages may still be stored with the account to support history, follow-up, safety and review under the approved India-only processing boundary.
An approved model service must not use learner messages, answers or performance history to train a general-purpose model. Personalisation for your requested explanation or strategy is not model training. Do not type passwords, OTPs, card details, UPI PINs, Aadhaar, PAN, health records, another person's confidential information or anything unnecessary for exam preparation.
The Privacy Policy explains storage, service providers, security, retention and how to make a request in more detail.
Back to top07Original content and review
Books, previous-paper files and other supplied research material are kept out of the question bank learners see. Transcribing, translating, lightly paraphrasing or changing names and numbers does not make copied material original. New questions and solutions must be independently written and pass rights and academic review before release.
An AI-generated draft is not published automatically. If AI helps write an item, it must still pass rights and academic review, be labelled as AI-assisted, and include a way to report a mistake.
Back to top08Safe and responsible use
AI features are limited to legitimate learning, practice, review and preparation planning. They may refuse or redirect requests that are unsafe, abusive, unlawful, unrelated to IBPS preparation, seek confidential live-exam answers, expose another person’s data or try to bypass safety controls.
Do not present an AI response as an official authority statement or use it to impersonate another person. Do not automate bulk requests, scrape responses or use the service to build a competing content bank. Good-faith research, accessibility use and reporting a safety or accuracy problem are welcome.
Back to top09Consent and learner controls
AI Usage consent is optional for creating an account and using non-AI learning. It is required only before Magic AI or PRGNA processes a learning request. We record which policy you accepted and when. If this policy changes materially, we may ask you to review and accept the update before affected AI use continues.
You can stop using AI without deleting the identity account. To ask about stored AI data, withdraw AI use or request eligible deletion, use Help → Privacy & consent after sign-in or email Support.scorevedaa@neuramach.ai. Withdrawing AI consent disables the affected AI processing; it does not silently cancel or purchase anything.
Back to top10Updates and contact
The version and update date appear at the top. We may update this policy when AI services, safety controls, available capabilities or data practices change. A material change will be shown clearly and may require you to accept the update before continuing.
Questions or suspected AI errors can be sent through the contact page or to Support.scorevedaa@neuramach.ai. The same inbox accepts privacy questions. ScoreVedaa IBPS is provided by NeuraMach AI Studio Private Limited.
Back to topNeed a plain-language answer?
Ask about this policy without sharing passwords, OTPs or payment credentials.