AI & Data Use

India's AI Governance Guidelines, released in November 2025, use a risk-based and proportionate approach built around trust, people-first design, fairness, accountability, understandability, safety and resilience. The Government has also stated that the framework relies on existing laws rather than creating a new horizontal AI statute at this stage.

Accordingly, AI use can still be subject to existing laws on personal data, information technology, consumer protection, intellectual property, sector regulation and other applicable obligations.

AI may be used to help users search, summarise, explain, compare or navigate source material and to support structured calculations or operational workflows. AI is treated as an assistance layer, not as an authority that can replace current source documents or professional judgment.

Knowledge pages that do not contain an interactive AI feature do not become AI services merely because AI may have assisted an internal drafting or review workflow.

For product-specific, safety-critical or compliance-sensitive questions, authoritative source material takes priority. Depending on the question, that may include the current product label, Safety Data Sheet, Technical Data Sheet, approved SOP, manufacturer instructions, current law, regulator guidance or another identified primary source.

If an AI answer conflicts with a current authoritative source, the authoritative source controls. AI should surface uncertainty or missing evidence rather than inventing unsupported facts.

People should remain able to review important inputs, assumptions and source material before acting. AI output should not be treated as a regulatory approval, laboratory result, legal opinion, medical instruction, safety certification or guarantee unless an identified competent authority or professional actually provides that status.

Higher-consequence uses require proportionately stronger review. Dosing, concentration, chemical compatibility, workplace safety, compliance, procurement and financial decisions should be independently checked before implementation.

Users should provide only the information reasonably necessary for the task. Do not submit passwords, access tokens, payment-card data, confidential customer lists, medical information or other sensitive material unless the particular service genuinely requires it and clearly explains the handling of that information.

Personal data used in an AI interaction is subject to the privacy notice and legal framework applicable to that service. The Knowledge Privacy Policy describes the static Knowledge site separately.

AI services must not be used as a reason to bypass legal protections for children or persons acting through lawful guardians. Where processing a child's personal data requires verifiable parental consent or another safeguard under applicable law, that requirement must be followed by the service that performs the processing.

Interactive services may also impose their own age or eligibility conditions in addition to legal requirements.

AI systems can produce inaccurate, incomplete, outdated or biased output even when the language sounds confident. Relevant limitations include gaps in source coverage, ambiguous user inputs, retrieval errors, calculation mistakes and model-generated statements that are not supported by evidence.

Where practical, important answers should identify the source used, separate fact from inference and make material uncertainty visible to the user.

For calculators and structured AI tools, users should verify units, quantities, concentrations, conversion assumptions, rounding and source values before relying on a result. A mathematically correct calculation can still be unsafe if the input value or product instruction is wrong.

When a tool can validate ranges or reject impossible inputs, those checks should be used as safeguards rather than as substitutes for source verification.

This page does not make a blanket promise about whether every connected AI provider uses data for model training, how long every provider retains prompts, or where every model is hosted. Those facts depend on the specific service and provider configuration.

Where an Alle's ClinX AI service is launched, its service-specific notice should describe material data flows, providers and retention practices that users need in order to make an informed choice.

AI-assisted output may reflect or refer to third-party material. Users should not assume that generated wording, images, code or other output is automatically free of copyright, trademark, confidentiality or other rights.

Open Knowledge material remains subject to its published licence. AI does not expand the licence for third-party content or for Alle's ClinX trademarks and brand assets.

Do not use AI features to attempt unauthorised access, obtain secrets, introduce malicious instructions, evade safety controls or manipulate systems in ways that violate law or service terms. Suspected security issues should be reported through the process described in Trust & Security.

If an AI-assisted answer appears inconsistent with an official Alle's ClinX document, report the page, tool, product or document involved so the issue can be reviewed against the source.

AI technology, providers and regulation can change quickly. We may update this page as features, applicable law, IndiaAI guidance or connected services evolve.

If a service-specific notice conflicts with this general page about a concrete AI interaction, the more specific notice controls for that interaction, subject always to mandatory law.