AI regulation in India is still at the draft stage, so you can build an AI chatbot or automation today. On 8 October 2026, IT minister Ashwini Vaishnaw said the government will publish a consultation paper on AI regulation within a month, with safety and deepfakes as the main focus. A consultation paper is a request for public comments, not a rule, so the sensible move is to build in a way that survives whatever comes next.
This post is for owners and founders who are about to hire a developer for an AI project. After reading it you will know what has been announced, what is still unknown, and which five habits to ask your developer for now.
What is happening with AI regulation in India right now?
Speaking in New Delhi on Thursday 8 October 2026, Vaishnaw asked his department to prepare a “regulation-focused consultation paper” on AI. He said the paper should be out in about a month, which points to early November 2026 (that date is our reading, not an official one).
He named four focus areas: AI safety, harms such as deepfakes, a human-first approach, and skilling. He also said AI cannot be handled by law alone and called for a “techno-legal” solution, meaning rules backed by technical controls. Most important for businesses, he said industry must take primary responsibility for cybersecurity, safety and user harm.
Some background helps. The India AI Governance Guidelines, released earlier in 2026, already take a principle-based approach built on trust and responsible innovation. They list misinformation, deepfakes and algorithmic bias as major risks, and suggest reviewing existing laws on liability, data protection and content authentication.
A separate report says MeitY (the IT ministry) has already consulted on amendments to the IT Rules, 2021 covering synthetically generated information, which is the official term for AI-made fake content.
Do I need to wait for the rules before building an AI project?
No. Nothing announced so far bans or licenses a business chatbot, a document reader, a forecasting model or an internal automation. The paper has not even been published, so there is nothing to comply with yet.
Waiting also has a cost. A six-month delay on an automation that saves your team hours every week is a real loss, while the likely rules are about safety and accountability, which good practice already covers.
The risk sits in the other direction. A project built with no record of what data it uses, no human check and no way to switch it off is hard to fix later. That is cheap to avoid at the start and expensive to add after launch.
What could AI regulation in India mean for my chatbot or automation?
Nobody outside MeitY knows the contents of the paper. What we can do is map the stated focus areas to the kinds of projects Indian businesses actually build. Treat the right-hand column as planning guidance, not legal advice.
Project type | Stated focus area that may touch it | What could be asked of you | What to do now |
Customer support chatbot | Safety, user harm | Clear disclosure that the user is talking to AI; a way to reach a human | Add an “AI assistant” label and a human handover |
Image or video generation for ads | Deepfakes, synthetic content | Labelling of AI-made media; proof of consent for any real person’s face | Label outputs, keep a log of prompts and sources |
Document reading (invoices, KYC, GST) | Safety, accountability | Accuracy records, human review for decisions that affect people | Keep a human check on low-confidence results |
Hiring or credit scoring | Human-first approach, bias | Explanations for decisions, bias checks | Avoid fully automatic rejections |
Internal forecasting or analytics | Lower risk | Likely little beyond data protection duties | Keep a simple model note |
Want a fixed number for your project? Send 3–4 lines on WhatsApp and get a written quote within 48 hours.
What should I build into an AI project now?
Five habits cost little at the start and make you ready for most rule changes.
Human handover. Any AI that talks to customers or affects a person should pass to a human on request or when it is unsure. This is the single most useful safeguard.
Clear disclosure. Tell users when they are dealing with AI. A one-line label is enough.
A log. Store what the model was asked, what it answered and which version of the model gave the answer. If a complaint arrives, you can show what happened.
A data note. Write down what data the AI uses, where it is stored and who can see it. This also helps with the DPDP Act (India’s data protection law) when personal data is involved.
An off switch. You should be able to pause the AI feature without taking down the whole app.
None of these needs a large budget. A good developer adds them during the build, not as a separate project.
Who is responsible if my AI makes a mistake?
Today, you are. If your chatbot gives a customer wrong information or your document reader misses a figure on an invoice, the customer deals with your business, not with the company that made the AI model. The minister’s remark that industry carries the main responsibility points the same way.
This is why we put the human check and the log first. They let you catch an error early, fix it and show what happened. Ask any developer you hire how the system behaves when the AI is unsure, because the answer tells you a lot about their experience.
Does AI regulation in India change the cost of an AI project?
Probably not by much for a first project. The five habits above are part of normal engineering and sit inside the fixed quote, not on top of it.
Costs rise when the use case itself is high-risk, for example automated decisions about loans or hiring. Those need review steps, testing and documentation that a simple support bot does not. For reference, we quote a complete app system at ₹2.5–10 lakh, and other Indian cost guides quote ₹2–25 lakh for apps. An AI-only pilot is priced separately AI pilot fixed price.
We cannot price rules that do not exist yet. What we can do is write the scope clearly so that a later change is a small, separately quoted item, not a rebuild. Our guide to app development cost in India shows how scope drives price.
What is a safe first step for an AI project under uncertain rules?
Start with a pilot, a small paid test on your own data. Every AI project at Ureka Technologies starts with a fixed-price AI pilot of about 4 weeks, with accuracy measured on your real data before you commit to a full build.
A pilot suits uncertain rules for three reasons. The spend is small, you learn whether AI is the right tool at all, and the logging, handover and data note are built in from week one. If the consultation paper later asks for something extra, you have a working base to adjust, not a large system to unpick.
We also tell clients when AI is the wrong tool. A rules-based form or a plain spreadsheet formula is often cheaper and safer, and there is no rule against choosing it.
If you want an AI system built for your business, our team handles AI and ML development end to end. You own the code, the data and the models from day one, which matters if a future rule asks you to show how your system works. For chatbots and assistants specifically, see our generative AI development work.
What should I do this month?
1. List every place you plan to use AI, and mark the ones that make or influence decisions about people.
2. Ask your developer how they will handle handover, logging and disclosure.
3. Read the consultation paper when it is published and send comments through MeitY if it affects you.
4. Start with a small pilot, not a large build.
Describe the task you want to automate and we will tell you whether AI fits. Describe the task on WhatsApp
Reply within a few hours on WhatsApp. Fixed written quote within 48 hours.
Sources
AI regulations: Consultation paper in a month; safety, deepfakes among key focus areas, says Vaishnaw, ThePrint (PTI), 8 October 2026
Govt coming out with paper on AI regulations next month: Ashwini Vaishnaw, The Hans India (IANS), 8 October 2026
India AI Regulation 2026: MeitY consultation paper on safety, deepfakes and startup compliance, CareerIndia, 8 October 2026
FAQ and FAQ schema
FAQ for the end of the post (4–6 questions, each answer 40–80 words). Paste the JSON- LD below into the page.
Is there an AI law in India in 2026?
India has no single AI law yet. On 8 October 2026 the IT minister said a consultation paper on AI regulation will be published within a month, and rules would follow later. Existing laws such as the IT Rules, 2021 and the DPDP Act still apply to AI products that handle user content or personal data.
When will India publish its AI regulation consultation paper?
The minister said it would come within a month of 8 October 2026, which points to early November 2026. No exact date has been announced. A consultation paper is a draft for public comments, so it is not yet binding, and final rules would come after feedback.
Can I start an AI chatbot project before the rules are out?
Yes. Nothing announced so far stops a business from building a chatbot, a document reader or an automation. Build in a human handover, a clear AI label, a log of answers and a data note now, and you will be ready for most likely requirements. Starting with a small pilot keeps the spend low.
Do I need to label AI-generated content in India?
The announcement did not set a labelling rule. MeitY has consulted on amendments to the IT Rules, 2021 covering synthetically generated information, so labelling may become a requirement. Labelling AI-made images, video and chat replies now costs almost nothing and avoids rework later.
Will AI regulation make AI projects more expensive?
For a first project such as a support bot or a document reader, probably only slightly, since logging, disclosure and human review are part of normal engineering. High-risk uses such as loan or hiring decisions may need extra testing and documentation. We cannot price rules that have not been published.
Who is responsible if my AI chatbot gives a wrong answer?
In practice your business is, because customers deal with you and not with the AI model maker. The minister also said industry must take primary responsibility for safety and user harm. Keep a log, add a human handover for uncertain answers and review replies regularly.