
India DPDP Act holds organisations responsible for how personal data is processed. When employees use AI tools on personal data, the organisation must assess the purpose, lawful basis, safeguards and vendor arrangements. HR should set an AI usage policy, train staff and prepare for the phased deadlines.
Employees now use generative AI to draft emails, summarise resumes, write performance notes and analyse complaints. When these tasks involve personal data and unapproved accounts, privacy and compliance risks follow. The Digital Personal Data Protection Act, 2023 (DPDP Act) and the DPDP Rules, 2025 commence in phases, so HR needs to prepare now.
Commencement notification G.S.R. 843(E), dated 13 November 2025, sets three phases.
|
Phase |
Commencement |
What starts |
|---|---|---|
|
Phase 1 |
13 Nov 2025 |
Definitions and Data Protection Board provisions |
|
Phase 2 |
13 Nov 2026 |
Section 6(9) (Consent Managers) and Section 27(1)(d) |
|
Phase 3 |
13 May 2027 |
Main provisions: grounds for processing, notice, consent, fiduciary obligations, individual rights, penalties |
The Rules have their own schedule. Rules 1, 2 and 17 to 21 commenced on publication, Rule 4 on 13 November 2026, and Rules 3, 5 to 16, 22 and 23 on 13 May 2027. Phase 2 covers Consent Managers only, so employer duties on consent and employee data start in Phase 3. Treat the window before May 2027 as preparation time.
Yes. Digital personal data of employees and candidates is covered, subject to the Act's scope and exemptions. HR handles names, contact details, Aadhaar and PAN, resumes, payroll and bank data, appraisals, leave and health information, and disciplinary records.
The Act does not ban AI, but using AI does not remove the duty to run a lawful, safeguarded process. If a recruiter uploads resumes to an AI tool, assess the purpose, lawful ground, the tool's data handling and the safeguards.
|
DPDP term |
Meaning |
HR example |
|---|---|---|
|
Data Fiduciary |
Decides the purpose and means of processing |
Employer |
|
Data Principal |
The individual the data relates to |
Employee or applicant |
|
Data Processor |
Processes data on the Fiduciary's behalf |
Payroll or HRMS provider |
Under Section 8, the employer stays responsible for processing done by a Processor. The same test applies to an external AI service: look at its real role, contract and data handling. The Act has no separate "sensitive data" category, but financial, identity and health data deserve stronger safeguards.
Using an AI tool does not create a lawful ground.
The Act does not ban workplace AI. Risk arises when personal data is processed without a clear purpose, lawful ground, safeguards or oversight.
|
Scenario |
What can go wrong |
Control |
|---|---|---|
|
Recruiter pastes resumes into a public AI tool |
Candidate data processed without basis, notice or safeguards |
Approved tools, anonymised profiles |
|
Manager summarises appraisal notes |
Confidential data exposed; biased or wrong summary |
Restricted access, manager review |
|
Support staff paste complaints |
Customer identifiers leave approved workflows |
Mask identifiers, role-based access |
|
HR drafts a disciplinary letter |
Confidential data exposed; unproven claims enter the record |
Templates with placeholders, HR or legal review |
|
Team uses an AI meeting assistant |
Unannounced capture of employee or customer data |
Approve the tool, disclose recording, set retention |
|
Employee leaves with work chats in a personal AI account |
Data stays outside company control |
Company accounts, AI step in offboarding |
HR works with legal, IT, security and procurement.
Do: use approved tools and company accounts; anonymise and share only what is needed; verify outputs; keep human review on people decisions; report accidental exposure immediately.
Don't: upload identifiable employee or candidate data, or payroll, medical and disciplinary records, to unapproved tools; use personal AI accounts for confidential work; bypass security controls; treat AI output as the sole basis for a decision; hide mistakes, because deleting a prompt may not remove what was processed.
Training turns policy into practice. Treat it as an ongoing programme, not a one-off.
Vinsys helps organisations build practical AI capabilities through role-based training that connects AI tools with real business workflows and responsible usage.
Involve IT, legal, procurement and security before approving any AI platform for employee data. Check:
Employees should report at once and not try to fix it alone.
HR coordinates and supports communication, IT security leads containment, and legal leads the regulatory assessment.
These phases are a recommended schedule, not statutory dates.
|
Timeline |
Focus |
Key activities |
|---|---|---|
|
Now to Dec 2026 |
Visibility and ownership |
Inventory HR data and AI tools, find unapproved AI, assign owners |
|
Dec 2026 to Jan 2027 |
Policies and governance |
AI policy, documented purposes and lawful grounds, notice review, vendor approval |
|
Feb to Mar 2027 |
Security and operations |
Access controls, retention, incident workflows, rights requests, drills |
|
Apr to 12 May 2027 |
Training and validation |
Role-based training, policy acknowledgement, vendor gap closure, readiness assessment |
|
13 May 2027 onwards |
Operational compliance |
Ongoing monitoring and review |
Track progress monthly through indicators such as tools inventoried, vendors vetted, employees trained, high-risk use cases reviewed and open gaps.
Responsibility stays with the organisation as Data Fiduciary, but the work is shared.
Form a cross-functional AI governance group. A Data Protection Officer is mandatory only for Significant Data Fiduciaries.
A 500-person firm finds recruiters, managers, finance and support staff using AI, some on personal accounts. It inventories usage, sets approved rules through a cross-functional group, reviews HR data processing and vendors, trains by role, and monitors training completion, approved-tool use, incidents and new tools. The lesson: blocking public tools is not enough. Understand real behaviour, create approved workflows, vet vendors and train people.
1. Does the DPDP Act apply to employee data?
Yes. It covers digital personal data of employees and candidates, subject to scope and exemptions. Employers generally act as Data Fiduciaries.
2. Can employees use ChatGPT or Claude at work?
Yes. There is no blanket ban, but organisations must control how personal data is processed. Staff should use approved tools and follow internal policy.
3. Can HR upload CVs to an AI tool?
Not to unapproved tools. First assess the purpose, lawful ground, vendor practices, safeguards and internal approvals.
4. Is employee consent always required?
No. Section 7(i) recognises legitimate uses for employment and safeguarding the employer. Where consent is relied on, it must meet the Act's standards.
5. Is an AI vendor a Data Processor?
Possibly, if it processes data on your behalf and instructions. Review the actual arrangement, not the vendor's label.
6. Can we use overseas AI tools?
There is no blanket prohibition. Section 16 allows restrictions on transfers to specified countries, and sector rules and contracts may also apply.
7. What if an employee shares personal data with an AI tool by mistake?
Start incident response with IT security and legal: assess the exposure, contain it, document it and decide on notifications.
8. What are the penalties?
The Schedule provides for up to ₹250 crore for failing to take reasonable security safeguards and up to ₹200 crore for failing to notify a breach, depending on the circumstances.
9. Is a Data Protection Officer mandatory?
Only for Significant Data Fiduciaries. Other organisations should still assign clear internal ownership.
10. When do the main obligations apply?
The main operational provisions commence on 13 May 2027. Some institutional provisions commenced earlier, so track each date.
The DPDP Act has made personal data protection part of everyday governance. As employees use generative AI for recruitment, communication and performance work, the goal is not to restrict AI but to set clear boundaries so people can benefit from it while employee, candidate and customer data stays protected. Begin preparing now for the main provisions due on 13 May 2027.
Build AI capabilities alongside data protection awareness. Vinsys supports organisations through role-based AI training that connects emerging technologies with practical business workflows and responsible usage, helping teams adopt AI with confidence and accountability.

Vinsys Top IT Corporate Training Company for 2025 . Vinsys is a globally recognized provider of a wide array of professional services designed to meet the diverse needs of organizations across the globe. We specialize in Technical & Business Training, IT Development & Software Solutions, Foreign Language Services, Digital Learning, Resourcing & Recruitment, and Consulting. Our unwavering commitment to excellence is evident through our ISO 9001, 27001, and CMMIDEV/3 certifications, which validate our exceptional standards. With a successful track record spanning over two decades, we have effectively served more than 4,000 organizations across the globe.