Is using ChatGPT DPDP compliant?
It can be. The DPDP Act doesn’t ban AI tools; it applies to the personal data you put into them. We find where personal data reaches ChatGPT, Copilot and other AI tools, and fix what doesn’t fit the Act.
Where personal data reaches AI
What the DPDP Act asks of AI tool use.
The Act doesn’t mention AI tools. Its duties apply to them like any other system that handles personal data. These duties start on 13 May 2027.
The Act covers what goes into AI tools
It applies to digital personal data processed in India, and to processing outside India that is linked to offering goods or services to people in India (Section 3). Personal data in a prompt, an uploaded file or a call transcript is still personal data.
You stay responsible
You remain responsible for personal data that a vendor processes on your behalf (Section 8(1)).
A contract with the AI vendor
If you use a data processor for an activity related to offering goods or services to people, you need a valid contract with it (Section 8(2)).
A lawful purpose
Personal data can be processed for a purpose the person consented to, or for a legitimate use the Act lists (Sections 4 to 7).
Security safeguards
You must take reasonable security safeguards to prevent a personal data breach. This covers processing that a data processor does on your behalf (Section 8(5)).
Deletion
You must erase personal data when consent is withdrawn or the purpose is no longer served, unless a law requires you to keep it. You must also make your processors erase it (Section 8(7)).
This page is general information, not legal advice.
Find and fix: what we check.
Personal data in AI prompts is usually spread across tools nobody listed. We find each one, then fix what doesn’t fit the Act.
AI tool inventory
What we checkWhich AI tools and AI features are in use, approved or not.
What we fixAn AI tool register, added to your data map.
Data in prompts and files
What we checkWhich personal data goes into each tool, whose it is, and why.
What we fixData-flow maps that include AI vendors.
Vendor terms
What we checkWhether each vendor acts on your behalf, what it keeps, for how long, and where.
What we fixTerms flagged for our partner law firm, and a processor register.
Account settings
What we checkWork accounts, single sign-on, retention and model-training settings.
What we fixApproved tools set up, with the settings recorded.
Masking
What we checkWhether personal data can be removed before it reaches an AI model's API.
What we fixA masking step that catches Indian identifiers, with logs.
Deletion
What we checkWhether you can find and delete a person's data in AI tool histories, logs and vector stores.
What we fixDeletion jobs that cover AI data stores, while keeping the logs the Rules require.
Staff policy
What we checkWhat staff may put into which tool.
What we fixAn AI use policy that matches the approved tools.
Three ways to fix it, without banning AI.
Most companies use a mix. For data that should never leave your infrastructure, see Own Your AI.
Approve and set up
Pick the AI tools staff may use, put them on work accounts with a contract, and record the settings.
Mask before it leaves
Remove or replace personal data before a prompt reaches an AI model's API, and log what was masked.
Run models privately
For the most sensitive data, run AI models on infrastructure you control, so prompts never reach a third-party AI vendor.
Do you need an AI privacy impact assessment?
Under the DPDP Act, only if you are designated a Significant Data Fiduciary. Our guide to Significant Data Fiduciary readiness explains what designation would mean.
Required only for Significant Data Fiduciaries
The Act asks for a periodic Data Protection Impact Assessment only from a Significant Data Fiduciary, a company the government notifies under Section 10 (Section 10(2)(c)). Rule 13 then requires an assessment and an audit once in every twelve months, with significant observations reported to the Board.
No company is designated yet
As of 20 November 2026, the government has not notified any Significant Data Fiduciary. Rule 13 starts on 13 May 2027, and applies only if you are designated.
Where AI comes in
Rule 13 also asks a Significant Data Fiduciary to check, with due diligence, that its technical measures, including algorithmic software, are not likely to pose a risk to the rights of the people whose data it processes. For AI features, that check sits naturally inside the assessment.
Everyone else: optional, but useful
If you are not designated, the Act doesn’t require one. A short written assessment of each AI feature, covering the data, the purpose, the vendor and the safeguards, is still a practical record of the security safeguards you took (Section 8(5)), and it answers many customer questionnaires.
This page is general information, not legal advice.
How it runs.
AI tools are part of every DPDP gap assessment we run. We can also look at them on their own.
- Step 1
Discover
Find AI tools from single sign-on, expense and browser data, plus a short staff survey.
- Step 2
Map
Trace which personal data reaches each tool, and on what terms.
- Step 3
Fix
Approve, set up, mask or replace each tool, and write the staff policy.
- Step 4
Evidence
Record the register, settings and contracts, and send vendor terms to our partner law firm.
Related DPDP services.
DPDP gap assessment
Every system that holds personal data, AI tools included, compared against the Act and Rules.
See the assessmentDPDP implementation
Consent, rights requests, retention and breach workflows, built into your systems.
See implementationOwn Your AI
Private AI models on infrastructure you control, for data that should not leave it.
See Own Your AIDPDP and AI tools FAQs.
Is using ChatGPT DPDP compliant?
It can be. The DPDP Act does not ban AI tools. It applies to the personal data you put into them. The main things you need are consent or a legitimate use for that data, a valid contract with an AI vendor that processes it for you, reasonable security safeguards, and a way to delete the data when it is no longer needed. These duties start on 13 May 2027.
Is an AI vendor a data processor under the DPDP Act?
If it processes personal data only on your behalf, the Act treats it as a data processor, and you stay responsible for that processing. If it also uses the data for its own purposes, it may be responsible in its own right. Where the work relates to offering goods or services, the Act requires a valid contract. Our partner law firm reviews the vendor terms.
Can employees paste customer data into ChatGPT?
We advise allowing it only in a tool your company has approved for that purpose, with a contract and settings you have checked against the Act. The customer's consent, or a legitimate use, must also cover that use of their data. Personal accounts are hard to bring within these duties: your company has no contract with the vendor, and cannot check or control what the vendor keeps.
Does masking personal data before a prompt help?
Yes. If personal data is removed or replaced before the prompt leaves your systems, less of it reaches the AI vendor. Test that the masking catches Indian identifiers such as Aadhaar numbers, PANs and phone numbers. Masked data that can still be linked back to a person remains personal data in your systems.
Do you give legal advice?
No. Our engineers do the technical work. Our partner law firm reviews it and gives you signed, written legal documentation.