Quick answer
AI hiring tools process candidates' personal data and inform decisions about them, so the DPDP Act applies in full from 13 May 2027. Employers need a lawful basis for each use — the application itself, talent pools, background checks, model training — accurate and complete data under section 8(3), Rule 6 safeguards, a contract with the ATS or AI vendor, and retention that respects the Rule 8(3) one-year minimum.
- A CV submitted for a specific role is data voluntarily provided for that purpose; reuse for other roles needs consent.
- Section 8(3) requires accurate, complete and consistent data where it is used for decisions about a candidate.
- Keep a human decision-maker for rejections and record how AI scores were used.
- Background checks and social-media screening need their own basis; the public-data exclusion is narrow.
- ATS and AI screening vendors are Data Processors under contract; you remain liable.
- If candidates or roles are in the EU, AI used for recruitment is high-risk under Annex III of the EU AI Act.
Who this is for: HR and talent acquisition teams, HR-tech and ATS vendors, Recruitment agencies, GCCs hiring at scale, Data Protection Officers
How does the DPDP Act apply to AI hiring tools?
Recruitment is one of the most data-intensive things an employer does: CVs, contact details, education and employment history, salary expectations, assessment results, interview recordings, references and background-check reports. Adding AI — resume parsing and ranking, chatbot screening, coding-test proctoring, video interview analysis, candidate matching across a talent pool — multiplies the processing and makes it drive decisions.
Under the DPDP Act:
- The employer deciding why and how candidate data is processed is the Data Fiduciary.
- The ATS, assessment platform or AI screening vendor is usually a Data Processor acting on the employer's instructions, under a contract required by section 8(2).
- A recruitment agency that sources candidates for multiple clients and keeps its own database is a Data Fiduciary for that database, and shares data with clients as a separate fiduciary or processor depending on the arrangement.
The Act has no specific article on automated decision-making, but section 8(3) applies directly: where personal data is likely to be used to make a decision that affects the Data Principal, the fiduciary must ensure its completeness, accuracy and consistency. Rejecting or shortlisting a candidate is such a decision.
All these obligations apply from 13 May 2027. Until then, SPDI Rules apply to any sensitive information collected, such as health or financial details.
Lawful basis for each hiring purpose
| Purpose | Likely basis | Notes |
|---|---|---|
| Assessing a candidate for the role she applied to | s.7(a) — voluntarily provided for a specified purpose — or consent | Notice should explain AI-assisted screening |
| Keeping her CV for future roles (talent pool) | Consent | Separate opt-in; set a retention period |
| Reference and background checks | Consent | Third-party verification vendors are processors or separate fiduciaries |
| Screening public social-media profiles | Narrow; consent preferred | s.3(c)(ii) covers only data she herself made public; inferences from it are new processing |
| Training or tuning your screening model on past applicants | Consent, or careful legal analysis | Separate purpose from assessing her application |
| Onboarding after an offer is accepted | s.7(i) employment purposes | From the employment relationship onwards |
Section 7(i) covers processing "for the purposes of employment". Whether it extends to pre-offer recruitment is a question of interpretation; the safer path is to rely on section 7(a) for the applied role and consent for anything broader.
Consent must be free, specific and limited to necessary data. Making a candidate agree to unrelated uses — marketing, data sharing with partners — as a condition of applying is unlikely to be valid.
Accuracy, fairness and human decisions
Section 8(3) turns AI quality issues into compliance issues. A resume parser that misreads dates, a ranking model that penalises career breaks, or a video tool that scores candidates on camera quality can produce inaccurate data driving decisions.
Controls that evidence compliance:
- Validate parsing. Let candidates review and correct extracted fields before submission; that also supports the section 12 right to correction.
- Explain the criteria. Score against job-related criteria the hiring team set, and record them.
- Test for disparate impact across gender, age bands, disability and other relevant groups before deployment and periodically after.
- Keep a human decision-maker. AI can rank and summarise; a person decides rejections, with the evidence visible.
- Avoid inference of characteristics that are not job-related — emotion, personality from facial expressions, health from voice.
- Provide accessible alternatives for candidates with disabilities who cannot use a timed or video-based assessment.
- Log scoring — model version, inputs used, score, reviewer and decision — and keep those logs for at least one year under Rule 8(3).
MeitY's India AI Governance Guidelines emphasise fairness and equity and understandable-by-design systems among their seven sutras; they are not binding, but they indicate the direction of policy.
Video interviews, proctoring and biometrics
Video interview platforms and online test proctoring often capture face images, voice, screen activity and sometimes keystroke patterns. These are personal data, and face templates or voiceprints are biometric — not a special DPDP category, but high-risk.
Guidance:
- Recording an interview for review by the hiring panel is generally within the application purpose if the candidate is told in advance.
- Automated analysis of facial expressions, tone or "engagement" is a separate, higher-risk use. Get specific consent, explain what is analysed, offer an alternative and consider whether it is job-related at all.
- Proctoring — webcam monitoring, room scans, screen capture — should be proportionate to the assessment stakes, with short retention for raw footage.
- Identity verification with a selfie and ID document should store only what is needed and delete images after verification.
For EU candidates, the EU AI Act's Article 50 requires deployers of emotion recognition or biometric categorisation systems to inform people exposed to them. See facial recognition and biometrics.
ATS vendors, security and retention
Vendor contracts
ATS, assessment and AI screening vendors are processors. Under section 8(2) and Rule 6(1)(f), contract for: processing only on your instructions; no use of your candidates' data to train the vendor's general models without authorisation; security safeguards; breach notice fast enough for your Rule 7 reporting; retention including the Rule 8(3) minimum; deletion on instruction; sub-processors; and support for rights requests. Ask how the vendor's model was trained and tested for bias. See AI vendor due diligence.
Security
Candidate databases are attractive targets. Apply Rule 6: encrypt CVs and assessment data, restrict access to the hiring team for each role, log access and exports, and back up securely.
Retention
- Unsuccessful candidates: erase when the recruitment purpose is served (section 8(7)), subject to the Rule 8(3) requirement to retain personal data and processing logs for at least one year from processing. Keep that copy restricted.
- Talent pool: only with consent, with a defined period and a renewal prompt.
- Hired candidates: move necessary data into the employee record under section 7(i), and delete the rest.
Candidate rights and the EU AI Act overlap
Rights
Candidates are Data Principals with rights under sections 11 to 14: access to a summary of their data and processing, correction and erasure, grievance redressal within your published period of no more than 90 days (Rule 14(3)), and nomination. Publish how to make requests on your careers page (Rule 14(1)), and include your contact person's details in every response (Rule 9). Expect candidates to ask why they were rejected; have a clear, human answer about the criteria used.
EU AI Act
If you hire in the EU, or your HR-tech product is placed on the EU market or its output is used in the EU, AI systems used for recruitment and selection — including targeted job ads, filtering applications and evaluating candidates — fall under Annex III high-risk use cases. Most high-risk obligations apply from 2 August 2026; the Commission's proposed Digital Omnibus may delay some deadlines, so check its adoption status. Non-EU providers of high-risk systems need an EU authorised representative. GDPR also applies to EU candidates' data.
See EU AI Act high-risk rules for HR tech and HR and employee data under DPDP.
Further reading on suhasbhairav.com
- BlogAI Recruiting Agents in Production: Conversation Screening vs Resume Rule Matching
- BlogCompliance testing for high-risk AI: governance
- Prompts100 Best ChatGPT Candidate Screening Prompts for HR and Recruiters
- AI Use CasesAI Use Case: Hr Interview Notes and Hiring Summaries
- AI AutomationAI Automation for Recruiting in Human Resources
Before 13 May 2027
Compliance checklist
Map every hiring data flow
List ATS, assessments, video tools, background checks and AI scoring with data, vendors and regions.
Section 8(1)
Update the candidate privacy notice
Itemise data collected, explain AI-assisted screening and link to rights, withdrawal and Board complaints.
Rule 3
Separate talent-pool and training consent
Ask specifically before keeping CVs for other roles or using them to train models.
Section 6
Let candidates review parsed data
Show extracted fields for confirmation to support accuracy and correction.
Sections 8(3), 12
Test screening models for bias
Measure outcomes across groups before deployment and on a regular schedule.
Section 8(3)
Keep humans on adverse decisions
Require a named reviewer for rejections and log the decision with the AI score.
Section 8(3)
Contract ATS and AI vendors
Cover instructions, no training without authorisation, safeguards, breach notice, retention and deletion.
Section 8(2); Rule 6(1)(f)
Set candidate retention rules
Erase unsuccessful candidates' data after the purpose, keeping only the Rule 8(3) one-year minimum in restricted storage.
Section 8(7); Rule 8(3)
Publish a candidate rights channel
Put request means, identifiers needed and the grievance period on the careers page.
Rule 14
Key dates
DPDP Rules notified
EU AI Act Annex III high-risk obligations apply (check Digital Omnibus status)
DPDP obligations apply to hiring data and AI screening
Answers
Frequently asked questions
Is AI resume screening legal in India?
Is AI resume screening legal in India?
Yes, if done under the DPDP Act's rules from 13 May 2027: a lawful basis for each use, a notice explaining AI-assisted screening, accurate data for decisions under section 8(3), security safeguards and a contract with the vendor. Keeping a human on rejection decisions is strongly advisable.
Do I need candidate consent to use AI in hiring?
Do I need candidate consent to use AI in hiring?
Not necessarily for assessing the role she applied to, which can rely on data voluntarily provided for that purpose under section 7(a), with a clear notice. Consent is generally needed for broader uses, such as talent pools, background checks, emotion analysis or training your screening model.
How long can employers keep rejected candidates' data?
How long can employers keep rejected candidates' data?
Only as long as the recruitment purpose needs, then erase under section 8(7). Rule 8(3) requires personal data and processing logs to be retained for at least one year from processing, so keep a restricted copy for that minimum. Talent-pool retention requires consent.
Can candidates ask why an AI rejected them?
Can candidates ask why an AI rejected them?
Candidates can request a summary of their personal data and processing under section 11, and raise grievances under section 13. The DPDP Act has no explicit right to an explanation, but section 8(3) accuracy duties and good practice favour a clear, human answer about the criteria used.
Is an ATS vendor a Data Processor under DPDP?
Is an ATS vendor a Data Processor under DPDP?
Usually yes, when it processes candidate data only on the employer's instructions. The employer remains the Data Fiduciary and responsible under section 8(1). If the vendor uses candidate data to train its own general models, it begins acting for its own purpose.
Does the EU AI Act apply to Indian HR-tech companies?
Does the EU AI Act apply to Indian HR-tech companies?
It can. AI used for recruitment and selection is high-risk under Annex III, and the Act reaches non-EU providers placing systems on the EU market or whose output is used in the EU. Most high-risk duties apply from 2 August 2026, subject to the proposed Digital Omnibus.
Legal references and sources
- Digital Personal Data Protection Act, 2023 — Sections 6, 7(a), 7(i), 8(3) (official text)
- Digital Personal Data Protection Rules, 2025 — Rule 6 (official text)
- Digital Personal Data Protection Rules, 2025 — Rule 8(3) (official text)
- EU AI Act — Annex III (employment) (official text)
- DPDP Rules, 2025 (G.S.R. 846(E)) ↗
- MeitY — Data protection framework ↗
- EU AI Act (Regulation (EU) 2024/1689) ↗
- IndiaAI ↗
This guide explains the law for engineering and business teams. It is not legal advice; confirm how it applies to your organisation with qualified counsel, and check the Gazette for later notifications or amendments.
About the author

Suhas Bhairav
Systems architect and applied AI practitioner who designs production AI systems for Indian and global companies, with a focus on privacy-by-design under the DPDP Act and the EU AI Act.
Related guides
- Sector guidesDPDP for HR: Employee Data and the Employment Legitimate UseWhat section 7(i) lets employers do without consent, where its limits are, and how HR handles candidates, monitoring, dependants, ex-employees and HR tech under DPDP.
- EU & cross-borderEU AI Act High-Risk Rules for Indian HR-Tech and Credit ScoringWhen Indian HR-tech and credit scoring products become high-risk AI under EU AI Act Annex III, what providers and deployers must do, the August 2026 date and the pending Digital Omnibus delay proposal.
- AI & personal dataDPDP Act for AI Systems: Implementation Guide for Indian CompaniesHow the DPDP Act and Rules apply to LLM apps, RAG, agents and ML pipelines in India, with the architecture controls to build before 13 May 2027.
- AI & personal dataFacial Recognition and Biometrics Under the DPDP ActFace recognition, fingerprint attendance and other biometrics are personal data under the DPDP Act. How to choose a basis, protect templates and avoid over-collection.
- AI & personal dataAI Vendor Due Diligence Checklist Under the DPDP ActA practical due diligence checklist for LLM APIs, vector databases, observability tools and AI SaaS vendors, mapped to DPDP section 8 and Rules 6, 7 and 8(3).
- DPDP Act 2023DPDP Legitimate Uses: Processing Without Consent Under Section 7Section 7 of the DPDP Act lists the only situations where personal data can be processed without consent. Here is each ground, its limits and how to document it.