AI Use Disclosure.

What this page is about

SMMARUN's commitment to AI transparency. Article 50 of the EU AI Act (Regulation (EU) 2024/1689) requires that anyone interacting with an AI system on a website be told, clearly and in good time, that they are doing so. The UNESCO Recommendation on the Ethics of AI 2021 (§§22-23) frames the same duty in ethical terms: transparency and explainability. The MeitY 7 Sutras (February 2026) name it as the Transparency Sutra. This page is how we honour all three on smmarun.ai.

To be plain about the current state: this site does not currently deploy a chatbot, a voice agent, or AI-assisted search. There is no automated conversational interface on smmarun.ai today. If and when we deploy one, the section below tells you what to expect, before you use it.

When AI interacts with you

Article 50 disclosure is triggered when an AI system is, or would be, the thing on the other side of an interaction with a natural person. On a website, that typically means:

• A chatbot answering questions in a chat window.
• A voice agent taking a call or rendering speech.
• AI-assisted search that does more than match keywords (for example, a system that summarises results or answers questions over our content).
• AI-generated summaries or recommendations rendered to you on the page, where the substance comes from a model rather than a person.

None of these are live on smmarun.ai today. We will not deploy any of them without first updating this page, posting a visible notice at the point of interaction, and offering a way to reach a human instead.

What we will tell you, when we deploy AI on this site

If we add a chatbot, a voice agent, AI-assisted search, or any other AI-mediated interaction to smmarun.ai, we will disclose the following at the point of interaction, before you engage:

• What it is. A plain-language statement that you are interacting with an AI system, not a person. Where relevant, we will name the underlying model provider (for example, Anthropic Claude).
• Its purpose. What it is for, and what questions it is designed to answer. What it is not designed to answer.
• Its limitations. Where it may be wrong, what it cannot do, and the categories of question it should not be relied on for (regulated advice, legal opinions, anything material to a business decision).
• Its training-data category. Whether it is a general-purpose large language model, a retrieval-augmented system grounded in our own published content, or something else. We will be specific about whether the system was trained, fine-tuned, or grounded on SMMARUN materials.
• The human-in-the-loop arrangement. Who reviews what, when, and on what cadence. Whether responses are pre-reviewed, post-reviewed, or unreviewed; and what happens if you flag a response as wrong.
• The channel to escalate to a human. A visible, working route to a named person, on the same page, before you commit to the interaction.

These six points are the minimum. We may add more depending on the system. We will not deploy an AI interaction on this site that does not meet all six.

AI in our back-office

Separately from this website, SMMARUN uses AI tools internally. Our partners and analysts use general-purpose large language models (currently Anthropic's Claude, and other comparable systems) to draft documents, summarise long materials, structure analysis, and support research. The final work product carries a SMMARUN authorial responsibility: a partner reviews, edits, and signs off before anything goes to a client or to this site.

We do not use AI to make automated decisions that have a legal or significantly similar effect on any individual. That posture aligns with the right against automated decision-making in DPDPA §10(d) and GDPR Article 22. If an automated decision is ever introduced anywhere in our service, we will disclose it specifically, name the decision, and provide a route to human review.

No training on your data

We do not pass form submissions, enquiry contents, gated-content requests, or any other personal data you submit to us into any AI provider's training pipeline. Our commercial agreements with Anthropic and other AI providers we use contractually prohibit training on our content, including any personal data inside it. The same prohibition extends to fine-tuning, evaluation sets, and reinforcement-learning feedback signals derived from our usage.

Inputs we send to a model in the course of internal drafting are processed for that single inference and discarded by the provider in line with the zero-retention or short-retention terms in our agreements. We list the relevant providers and their roles in the sub-processor table on our Privacy Notice.

AI-generated content on this site

Some of the content on smmarun.ai is drafted with the help of AI tools. Where that is the case, the work has been edited, fact-checked, and signed off by a named SMMARUN author before publication; the canonical authorial responsibility rests with SMMARUN and the named partner, not with the model.

Where a particular page or artefact is materially AI-generated (for example, an automated summary, a model-produced chart, or a synthetic illustration), we will mark it inline at the point of reading. Where a page is fully human, no disclosure is needed: that is the baseline.

We do not publish content under fabricated bylines, and we do not pass off model output as the verbatim view of a named SMMARUN partner without their review and consent.

How to opt out of AI interactions

If, in the future, an AI-mediated interaction appears on this site and you would rather reach a human directly, you can do so without engaging the AI at all. The escalation route will be visible at the point of interaction. You can also write to us at any time:

Human-only contact Email: hello@smmarun.ai with the subject Human contact request. Postal: SMMARUN Management Consortium Associates LLP, Gurugram, Haryana, India.

We will respond from a named SMMARUN colleague, not from an automated system. Opting out of AI interactions does not affect any other right you hold under the DPDPA, the GDPR, or any other applicable law.

Changes to this disclosure

We may update this page from time to time, particularly if we introduce a new AI-mediated interaction or change an existing one. Material changes will be flagged on the site at the top of the next visit, and the Last reviewed date below will be updated. Previous versions are available on request.