AI Hiring in India: Can Companies Reject Candidates Using Algorithms?
Introduction
You apply for a job online.
Within seconds, your application is rejected.
No recruiter calls you.
No human interviews you.
The company uses software that scores:
- your CV;
- employment gaps;
- university;
- previous employers;
- psychometric answers;
- recorded video interview;
- facial expressions;
- voice patterns;
- typing behaviour;
- and even the words you use.
An algorithm decides that you are not a “good fit.”
Can a company legally reject a candidate this way in India?
The answer in 2026 is:
AI-assisted recruitment is not illegal in India merely because an algorithm helps screen or rank candidates. But employers cannot assume that using AI gives them immunity from discrimination, privacy, employment-law or constitutional obligations.
India currently has no single dedicated statute specifically regulating AI recruitment systems.
Instead, AI hiring is governed through a combination of:
- existing employment laws;
- anti-discrimination provisions;
- constitutional equality protections in public employment;
- privacy and data-protection law;
- sector-specific laws;
- and India’s emerging AI-governance framework.
India’s AI Governance Guidelines, released in November 2025 and reinforced through the 2026 national AI-governance framework, expressly identify concerns such as:
- bias;
- discrimination;
- unfair outcomes;
- exclusion;
- opacity;
- and lack of accountability.
The Government’s approach is currently to use existing law and sectoral regulation rather than enact one horizontal AI statute for every AI application.
The most important legal principle is therefore:
An employer cannot avoid responsibility for an unlawful hiring decision simply by saying, “the algorithm made the decision.”
The Short Answer
Yes, employers can use AI to assist recruitment.
AI may be used for:
- CV screening;
- ranking applicants;
- skills matching;
- interview scheduling;
- candidate assessments;
- video interviews;
- fraud detection;
- recruitment analytics.
But legal risk increases where the AI system:
- discriminates against protected groups;
- uses irrelevant personal characteristics;
- relies on inaccurate data;
- makes opaque decisions that cannot be justified;
- collects excessive personal information;
- analyses biometric or behavioural data unnecessarily;
- or automatically rejects candidates without meaningful oversight in high-impact situations.
Public-sector recruitment faces particularly strict scrutiny because Articles 14 and 16 of the Constitution require fairness, non-arbitrariness and equality of opportunity in public employment.
Private-sector employers also face statutory restrictions in specific areas, although India does not yet have one comprehensive private-employment discrimination statute covering every protected characteristic.
What Is AI Hiring?
AI hiring refers broadly to the use of automated or machine-learning systems in recruitment.
Examples include software that:
- scans CVs for keywords;
- assigns candidate scores;
- predicts job performance;
- ranks applicants;
- analyses online tests;
- compares applicant profiles;
- evaluates recorded interviews;
- recommends who should proceed;
- automatically rejects applicants.
Some systems merely help recruiters.
Others effectively make the decision.
That distinction matters.
AI-Assisted vs Fully Automated Hiring
AI-assisted hiring
The software recommends:
Candidate A: 87
Candidate B: 78
Candidate C: 42
A human recruiter then reviews the candidates and makes the final decision.
Fully automated hiring
The software decides:
Score below 60 → reject automatically.
No human meaningfully reviews the candidate.
The second model creates much greater legal and governance risk because errors or discriminatory patterns can operate at scale without correction.
Is Automated Rejection Automatically Illegal?
No.
There is currently no general Indian law saying:
“Every recruitment rejection must be made personally by a human.”
Companies can automate routine screening.
For example, an employer hiring a chartered accountant may reasonably configure software to reject applicants who do not possess a mandatory qualification.
Likewise, recruitment software may filter applications based on:
- location;
- qualification;
- experience;
- statutory licence;
- role-specific certification.
The problem arises where the algorithm uses criteria that are:
- unlawful;
- discriminatory;
- irrelevant;
- inaccurate;
- or impossible to justify.
Example: Legitimate Automated Screening
A hospital advertises for a role requiring:
- MBBS qualification;
- valid medical registration;
- five years’ experience.
The recruitment system rejects an applicant who lacks the required licence.
That is ordinary eligibility screening.
The fact that software performed the check does not itself make the rejection unfair.
Example: Potentially Discriminatory Screening
The AI system discovers historically that most successful senior employees were men aged 28–35.
It therefore learns to downgrade:
- women;
- older applicants;
- candidates with maternity-related career gaps.
The software may never contain a rule saying:
“Reject women.”
Yet its output can still produce discriminatory consequences.
This is one of the central risks of algorithmic hiring.
Algorithms Can Reproduce Historical Bias
AI systems commonly learn patterns from historical data.
Suppose a company historically hired:
- mostly men;
- mostly graduates from five elite colleges;
- mostly candidates from large metropolitan cities.
The algorithm may learn that those characteristics predict “success.”
But those patterns may reflect historical recruitment choices rather than genuine ability.
The system can then perpetuate the same exclusion automatically.
India’s AI Governance Guidelines specifically recognise bias, discrimination, exclusion and unfair outcomes as AI risks requiring safeguards.
India’s AI Governance Guidelines
India unveiled its AI Governance Guidelines in November 2025.
The framework uses seven broad principles, including ideas centred on:
- people-first AI;
- fairness and equity;
- accountability;
- understandable systems;
- trust;
- safety;
- responsible innovation.
The Government has stated that the model is risk-based, evidence-led and proportionate, and that high-risk systems should not be treated as unrestricted simply because no dedicated AI Act exists.
The guidelines are not themselves equivalent to a comprehensive AI statute creating a standalone cause of action for every hiring dispute.
But they are highly relevant to what responsible AI deployment should look like.
India Has Chosen Existing Law Plus AI Governance
The Government has expressly stated that it currently prefers using:
- existing legislation;
- sectoral regulation;
- technical safeguards;
- institutional coordination
rather than immediately creating one horizontal AI law covering every AI system.
In 2026, the Government also constituted:
- the AI Governance and Economic Group;
- and the Technology and Policy Expert Committee
to coordinate AI governance and advise on legal, technical and policy issues.
This means the legal framework is developing rapidly.
Public-Sector Hiring Is Different
If AI is used by:
- Government departments;
- public authorities;
- State instrumentalities;
- public-sector institutions
for recruitment, constitutional law becomes extremely important.
Article 16 guarantees equality of opportunity in matters relating to employment or appointment under the State.
Article 14 prohibits arbitrary State action.
Therefore, a Government recruitment algorithm that produces arbitrary or discriminatory results can potentially face constitutional judicial review.
Public Employer Cannot Say “The Computer Decided”
Suppose a public authority uses an algorithm to rank applicants for Government jobs.
The model systematically disadvantages women without any legitimate job-related reason.
The Government cannot necessarily defend the process simply by saying:
“The ranking was automatically generated.”
The State remains responsible for the recruitment process it adopts.
Technology is a tool used by the decision-maker.
It is not a separate constitutional actor absorbing liability.
Supreme Court and Systemic Recruitment Discrimination
The Supreme Court’s recent jurisprudence has also recognised the danger of systemic discrimination in recruitment processes.
In a 2025 decision, the Court discussed how apparently neutral recruitment procedures can produce discriminatory outcomes and referred to systemic discrimination that arises through established recruitment and hiring processes even where discrimination is not openly stated.
This principle is highly relevant to algorithmic recruitment.
An AI model need not explicitly contain a discriminatory command to produce discriminatory effects.
Private Companies: Are Constitutional Equality Rules Directly Applicable?
Usually not in the same manner.
Article 16 specifically concerns employment or appointment under the State.
A purely private employer is therefore not automatically subject to the entire constitutional public-employment framework in the same way as a Government employer.
However, private companies are not operating in a legal vacuum.
Various statutes restrict discrimination in particular contexts.
Gender Discrimination in Recruitment
The Code on Wages, 2019 prohibits discrimination on the ground of gender in specified wage and employment contexts.
Section 3 also prohibits discrimination on the ground of sex while recruiting employees for the same work or work of similar nature, except where employment of women is lawfully prohibited or restricted.
The Labour Ministry has stated that the protection includes transgender persons in relation to:
- wages;
- recruitment;
- and employment conditions
for the same or similar work.
An AI recruitment tool cannot lawfully be used to bypass these requirements.
Example: Algorithm Downgrades Women
Suppose two applicants have equivalent qualifications.
The AI system systematically lowers scores for women because historical company data shows fewer women in senior technical roles.
If that causes discriminatory recruitment decisions, the employer cannot simply defend itself by saying:
“Gender was never entered manually.”
A discriminatory outcome caused by a model can still create legal concerns.
Transgender Applicants
The Transgender Persons (Protection of Rights) Act, 2019 expressly prohibits discrimination against transgender persons in employment and occupation.
The Act applies to establishments and prohibits unfair treatment in employment as well as denial or termination of employment on prohibited grounds.
Therefore, an automated recruitment system that filters out transgender applicants can expose the employer to statutory risk.
HIV Status
The HIV and AIDS (Prevention and Control) Act, 2017 prohibits discrimination including denial or unfair treatment in employment or occupation on the protected ground, subject to narrowly defined statutory exceptions.
An AI model should therefore not be configured to exclude candidates based on HIV-related information in circumstances prohibited by the Act.
Disability and AI Recruitment
Disability raises additional concerns.
The Rights of Persons with Disabilities Act, 2016 requires establishments to adopt equal-opportunity policies and contains significant employment protections and accessibility obligations.
AI recruitment tools can inadvertently disadvantage persons with disabilities.
For example:
- speech-analysis software may penalise a speech impairment;
- eye-contact scoring may disadvantage autistic candidates;
- timed online tests may disadvantage candidates requiring accommodation;
- facial analysis may fail for particular disabilities.
Employers should therefore examine whether the assessment tool itself creates barriers unrelated to actual job performance.
AI Video Interviews Are Particularly Risky
Some hiring platforms claim to analyse:
- tone of voice;
- facial movements;
- eye contact;
- pauses;
- emotion;
- confidence;
- vocabulary;
- speaking speed.
These tools can appear scientifically precise.
But employers should ask:
- Is the characteristic genuinely job-related?
- Has the tool been validated?
- Does it disadvantage disability groups?
- Does language or accent affect the score?
- Does lighting or camera quality matter?
- Can the vendor explain the model?
A candidate should not be rejected merely because an unreliable system interprets:
“limited eye contact”
as:
“low leadership potential.”
Accent Bias
Imagine an AI interview system trained largely on English speakers from one region.
It scores candidates partly on pronunciation and speech patterns.
Applicants with:
- Indian-language accents;
- regional accents;
- speech impairments
receive systematically lower ratings.
Unless accent is genuinely relevant to the role, such scoring can be difficult to justify as a fair measure of competence.
Career Gaps
AI can also misunderstand employment gaps.
A system may learn:
uninterrupted employment = high-performing candidate.
This may indirectly disadvantage people whose careers were interrupted due to:
- pregnancy;
- childcare;
- disability;
- illness;
- family caregiving;
- military service;
- further education.
A career gap should not automatically become a proxy for poor suitability.
College and Pincode Bias
AI recruitment systems may use proxy variables.
Suppose the model scores candidates based on:
- college attended;
- residential pincode;
- previous salary;
- English vocabulary.
These characteristics may correlate with:
- caste;
- class;
- geography;
- gender;
- socio-economic background.
Even where the algorithm does not directly receive protected information, proxy variables can reproduce structural exclusion.
AI Does Not Need to Know Caste to Produce Caste-Like Outcomes
This is a fundamental AI-governance issue.
Algorithms detect correlations.
A model may discover that candidates from particular:
- schools;
- locations;
- surnames;
- educational patterns
historically had certain hiring outcomes.
Using such correlations blindly can reproduce social bias.
Therefore, employers should audit not only direct variables but also proxies.
Can a Private Employer Reject Candidates for Any Reason?
No.
Private employment does provide employers substantial discretion in candidate selection.
But that discretion remains subject to:
- statutory discrimination restrictions;
- contractual representations;
- sectoral rules;
- privacy obligations;
- fraud/misrepresentation principles;
- and other applicable law.
There is no universal Indian statute requiring every private employer to prove that the “best” candidate was hired.
But there is a major difference between:
choosing Candidate A because they have stronger experience
and:
rejecting Candidate B because an algorithm unlawfully discriminates against them.
Does a Candidate Have a Right to Know Why They Were Rejected?
There is currently no universal rule requiring every private employer in India to provide every unsuccessful candidate with a detailed written explanation.
A company may therefore lawfully send:
“We have decided not to proceed with your application.”
However, lack of a general explanation requirement does not make unlawful discrimination lawful.
In public-sector recruitment, transparency and review obligations can be substantially stronger because the recruitment process is subject to constitutional and administrative-law principles.
Should Companies Tell Candidates AI Is Being Used?
As a governance matter, this is strongly advisable.
Candidates should ideally know when AI materially influences:
- screening;
- interview scoring;
- ranking;
- selection.
Transparency allows candidates to understand:
- what information is being analysed;
- what the tool is used for;
- whether human review exists.
India’s AI Governance Guidelines promote systems that are understandable and accountable rather than unnecessarily opaque.
Data Protection and Recruitment
AI hiring depends heavily on personal data.
A recruitment system may collect:
- name;
- email;
- phone number;
- CV;
- education;
- work history;
- address;
- salary history;
- photo;
- video;
- voice;
- test results;
- behavioural information.
India’s Digital Personal Data Protection Act, 2023 will therefore become increasingly important to recruitment technology.
But the DPDP Timing Must Be Stated Correctly
As of October 2026, many of the DPDP Act’s substantive provisions are not yet operative.
The Government’s 13 November 2025 commencement notification provides that core provisions including:
- Sections 3–5;
- most of Section 6;
- Sections 7–17;
- and various enforcement provisions
will commence 18 months after publication of the notification.
Similarly, major operational provisions of the DPDP Rules, 2025 become effective after the prescribed transition periods.
Therefore, it would be inaccurate in October 2026 to write:
“Every AI hiring platform is already fully regulated under all DPDP obligations.”
The framework is enacted and transitioning toward full operation.
What Will DPDP Mean for Recruitment?
Once the relevant provisions become operative, recruitment data processing will need to be examined through concepts such as:
- lawful processing;
- notice;
- purpose;
- security safeguards;
- Data Principal rights;
- retention;
- processor accountability.
Employers and recruitment vendors should therefore already be preparing their systems.
Can Companies Scrape Candidates’ Social Media?
This raises substantial privacy and fairness questions.
Recruitment tools may search publicly available:
- LinkedIn;
- GitHub;
- professional websites.
That can be relevant to a role.
But companies should be much more cautious about collecting unrelated information about:
- religion;
- political views;
- family;
- health;
- relationships;
- private lifestyle.
The fact that a piece of information can technically be found online does not automatically make it appropriate for employment scoring.
Social-Media Scoring Can Introduce Bias
Suppose an algorithm analyses candidate social profiles.
It infers:
- gender;
- religion;
- age;
- disability;
- political opinions.
The hiring team says:
“We never asked for these characteristics.”
But the AI derived them automatically.
This can create exactly the kind of hidden discrimination that makes automated recruitment difficult to audit.
Can AI Analyse Candidate Photographs?
A company should ask why it needs to.
Using a photograph merely to identify the interviewee is one thing.
Using facial features to predict:
- honesty;
- intelligence;
- loyalty;
- personality;
- leadership potential
is much harder to justify scientifically or legally.
High-impact decisions should not depend on unsupported pseudo-scientific inferences.
Biometric and Behavioural Data
Video-recruitment platforms may process highly sensitive categories of information in practical terms, including:
- face;
- voice;
- movement;
- behaviour.
Even where Indian law does not yet impose a complete sector-specific AI-hiring regime, organisations should adopt strict controls around:
- collection;
- access;
- storage;
- vendor use;
- deletion.
Can AI Reject a Candidate Because of Age?
Age discrimination in private recruitment does not currently have the same broad universal statutory regime as some foreign jurisdictions.
Nevertheless, age-based filtering can still create legal risk depending on:
- public/private employer status;
- applicable service rules;
- specific legislation;
- contractual representations;
- sector.
In public employment, age criteria must be supported by applicable recruitment rules and constitutional standards.
Employers should avoid allowing an AI system to invent its own age preference beyond legitimate job requirements.
AI Hiring in Government Jobs
Public recruitment presents the clearest legal accountability framework.
Imagine a State Government uses AI to shortlist applicants.
The software:
- secretly uses gender;
- changes weighting mid-process;
- cannot explain why candidates were rejected;
- produces inconsistent scores.
Such a system could potentially be challenged as:
- arbitrary;
- discriminatory;
- contrary to recruitment rules;
- violative of Articles 14 and 16.
Can Courts Review Algorithmic Decisions?
Yes, particularly where a public authority uses the system.
Judicial review can examine:
- legality;
- arbitrariness;
- discrimination;
- procedural fairness;
- compliance with statutory rules.
A Government cannot shield an unconstitutional decision behind proprietary software.
“The Model Is a Trade Secret” Is Not Always Enough
AI vendors frequently argue that the model is proprietary.
Commercial confidentiality can be legitimate.
But where software makes high-impact decisions affecting legal rights, complete opacity can become problematic.
Responsible deployment should allow meaningful understanding of:
- relevant factors;
- decision process;
- audit results;
- error rates.
India’s AI Governance Guidelines specifically emphasise understandable and accountable AI.
Human Review
Human oversight is one of the strongest safeguards employers can adopt.
But merely placing a human name on the workflow is not enough.
If the recruiter simply clicks:
Accept AI recommendation
for every candidate, human review is only cosmetic.
Meaningful human review should allow the recruiter to:
- question the score;
- examine evidence;
- override errors;
- consider context.
Automation Bias
Humans frequently trust algorithmic scores because computers appear objective.
This is called automation bias.
A recruiter may think:
“The software gave her 31/100, so there must be a reason.”
But the score may reflect:
- bad training data;
- broken parsing;
- proxy discrimination;
- incorrect CV extraction.
Employers should train recruiters to treat AI output as evidence to assess—not unquestionable truth.
Example: CV Parsing Error
Applicant:
8 years’ Java experience.
The PDF formatting prevents the recruitment software from reading that line.
The algorithm records:
Java experience: 0 years.
Candidate is automatically rejected.
This is not discrimination, but it illustrates why automated systems require:
- accuracy checks;
- appeal mechanisms;
- human review.
Example: Maternity Gap
An AI model gives lower employability scores to applicants with employment gaps longer than one year.
Many women with maternity/caregiving breaks are disproportionately excluded.
Even though the model never sees the word “woman,” its rule may produce gender-related effects.
Employers should test for this.
Example: Disability
A candidate with a speech disability completes an automated video interview.
The AI scores:
- fluency low;
- confidence low;
- communication low.
No human reviews the video.
The candidate is rejected from a role where spoken fluency is not essential.
This is a serious governance failure.
Example: Public Recruitment Algorithm
A public-sector employer ranks applicants using an algorithm that nobody can explain.
Candidates with higher objective exam scores are consistently placed below lower-scoring candidates.
In public employment, this could trigger constitutional scrutiny under Articles 14 and 16.
Example: Legitimate Coding Assessment
A software company asks applicants to complete the same coding problem.
AI checks whether:
- code runs;
- required test cases pass.
A recruiter reviews final results.
This is considerably easier to defend because the assessment is:
- role-related;
- consistent;
- measurable.
Background Checks
AI may also automate:
- education verification;
- employment verification;
- identity checks.
Companies should ensure the system distinguishes:
- genuine discrepancy;
- minor spelling variation;
- name change;
- clerical error.
Automatically treating every inconsistency as dishonesty can produce unfair outcomes.
Criminal Background Checks
Employers should be especially careful with automated criminal-history screening.
The system should distinguish:
- accusation;
- arrest;
- pending case;
- conviction;
- acquittal.
Treating an FIR as equivalent to conviction can be deeply misleading.
Role-specific legal requirements should guide the process rather than simplistic AI labels such as:
“High-risk candidate.”
AI Personality Tests
Many companies use automated psychometric assessments.
These may assess:
- personality;
- risk tolerance;
- leadership;
- teamwork.
Such tools are not automatically unlawful.
But employers should ask:
- Is the test validated?
- Is it relevant to the job?
- Does it unfairly exclude neurodivergent candidates?
- Is the score treated as absolute?
Emotion Recognition
Emotion-recognition tools deserve particular caution.
Claims that facial or vocal analysis can accurately determine:
- honesty;
- enthusiasm;
- integrity;
- emotional suitability
remain scientifically controversial.
Companies should not assume that sophisticated graphics make weak inference reliable.
Candidate Data Retention
Companies often retain rejected candidate profiles for future roles.
That may be commercially useful.
But indefinite retention creates privacy risk.
A defensible recruitment-data policy should specify:
- retention duration;
- purpose;
- deletion process;
- security.
This will become even more important as DPDP provisions fully commence.
Recruitment Agencies and AI Vendors
Employers frequently outsource AI hiring.
But outsourcing does not remove responsibility.
The employer should know:
- what data vendor collects;
- how scores are generated;
- whether data trains other models;
- where data is stored;
- whether vendor performs bias testing;
- whether candidates can be deleted.
Simply buying software from a famous vendor is not a compliance programme.
Vendor Contract Checklist
Employers should consider requiring:
- defined permitted purpose;
- confidentiality;
- security standards;
- deletion requirements;
- audit rights;
- bias testing;
- incident reporting;
- limits on model training;
- subcontractor controls;
- human-review mechanisms.
Should Candidates Be Allowed to Challenge a Score?
Even where not universally mandated today, this is good governance for high-impact decisions.
An appeal mechanism could allow a candidate to say:
“Your system incorrectly recorded that I do not have the required qualification.”
The employer can then correct obvious errors without litigation.
Can Candidates Sue Because They Were Rejected by AI?
Potentially, but the claim requires a legal basis.
Depending on the facts, possible grounds may involve:
- statutory discrimination;
- constitutional challenge for public employment;
- privacy/data issues;
- violation of recruitment rules;
- contractual or promissory claims;
- sector-specific protections.
The mere statement:
“AI rejected me”
is not by itself a complete legal cause of action.
The key question is:
What was unlawful about the decision or process?
Evidence Is Often the Hardest Problem
A candidate may suspect algorithmic discrimination but not know:
- which software was used;
- what variables it considered;
- what score they received;
- how others were scored.
This opacity is one of the biggest practical concerns in automated hiring.
It is also why organisations should maintain internal records and audits.
Bias Audit
A useful AI hiring audit should test whether selection rates materially differ across relevant groups.
Companies can examine:
- male/female selection;
- disability outcomes;
- regional outcomes;
- language groups;
- age bands where legally relevant.
A disparity does not automatically prove illegality.
But it can reveal a system that requires investigation.
Accuracy Audit
Employers should also test:
- CV parsing;
- qualification extraction;
- experience calculation;
- name matching;
- fraud flags.
A system can be unbiased but still be inaccurate.
Both problems matter.
Explainability
Recruiters should ideally be able to answer:
Why was Candidate X downgraded?
A response such as:
“Because the AI says so”
is poor governance.
A better system might show:
- mandatory qualification missing;
- experience below threshold;
- required test score not achieved.
Explainability improves:
- fairness;
- auditability;
- defensibility.
AI Hiring Compliance Checklist for Employers
Before deploying AI recruitment technology, ask:
- What decision does the system make?
- Is it advisory or automatic?
- What candidate data is collected?
- Are all variables relevant to the role?
- Could any variable act as a discrimination proxy?
- Has bias been tested?
- Has accuracy been tested?
- Does the system affect disabled candidates unfairly?
- Are candidates informed about AI use?
- Is there meaningful human review?
- Can errors be corrected?
- Can candidates challenge inaccurate information?
- How long is candidate data retained?
- Can the vendor reuse candidate data?
- Is biometric data actually necessary?
- Are public-sector constitutional requirements involved?
- Are Code on Wages gender protections satisfied?
- Are transgender/HIV/disability protections addressed?
- Are recruitment rules consistently followed?
- Is there an audit trail?
Candidate Checklist
If you suspect AI affected your application:
- preserve the job advertisement;
- save the eligibility criteria;
- screenshot automated interview instructions;
- save rejection communications;
- note whether AI/software was disclosed;
- preserve assessment results;
- request correction of obvious factual errors;
- document any indication of discriminatory treatment.
For public recruitment, also preserve:
- recruitment rules;
- merit lists;
- cut-offs;
- selection criteria;
- official notices.
Common Myths
Myth 1: “Using AI to recruit employees is illegal in India.”
Incorrect.
There is no general prohibition on AI-assisted recruitment.
Myth 2: “If an algorithm rejects the applicant, the company has no liability.”
Incorrect.
The employer remains responsible for the recruitment system it deploys.
Myth 3: “Algorithms cannot discriminate because computers are neutral.”
Incorrect.
AI can reproduce bias in:
- training data;
- variables;
- proxies;
- design.
India’s AI Governance Guidelines expressly recognise bias and discriminatory outcomes as material AI risks.
Myth 4: “Constitutional equality rules apply identically to every private company.”
Incorrect.
Articles 14 and 16 have particular importance in State/public employment. Private-sector liability must be analysed through applicable statutes and other law.
Myth 5: “Private employers can discriminate however they like.”
Incorrect.
Statutes restrict discrimination in areas including gender, transgender status and HIV-related employment discrimination.
Myth 6: “The DPDP Act is already fully applicable to every hiring algorithm.”
Incorrect as of October 2026.
Many substantive DPDP provisions commence after the statutory 18-month transition period from November 2025.
Myth 7: “Human review solves every AI problem.”
Incorrect.
Human review must be meaningful, not automatic rubber-stamping.
Myth 8: “A candidate always has a legal right to a detailed explanation.”
Not as a universal private-sector rule today.
Public recruitment and specific statutory contexts can involve stronger transparency obligations.
Frequently Asked Questions
Can companies use AI to reject candidates in India?
Yes, AI can be used for recruitment screening, but the employer remains responsible for compliance with applicable discrimination, employment, privacy and other laws.
Is fully automated hiring banned?
No general Indian ban currently exists.
However, fully automated high-impact decisions create substantially higher legal and governance risk.
Can AI reject women?
A company cannot use AI to circumvent applicable gender-discrimination protections. The Code on Wages regulates gender discrimination in recruitment for the same or similar work.
Can AI reject transgender candidates?
Discrimination against transgender persons in employment is expressly prohibited under the Transgender Persons (Protection of Rights) Act.
Can AI analyse video interviews?
Technically yes, but the employer should ensure that the analysis is:
- job-related;
- reliable;
- non-discriminatory;
- proportionate.
Can employers analyse facial expressions?
There is no blanket statutory prohibition today, but using questionable facial-emotion inference for hiring can create significant fairness, privacy and discrimination concerns.
Can AI use my social media?
Employers should carefully justify relevance and avoid unnecessary collection or inference of sensitive, unrelated information.
Does India have an AI Act like the European Union?
India’s current approach is different. The Government has adopted a principle-based AI Governance Framework and relies substantially on existing laws and sectoral regulation rather than one broad horizontal AI statute.
Can AI be used in Government recruitment?
Potentially, but the process remains subject to constitutional equality, fairness and recruitment rules under Articles 14 and 16.
Can I challenge an automated Government recruitment decision?
Potentially, especially where the system produces:
- arbitrariness;
- discrimination;
- rule violations;
- or procedural unfairness.
Is candidate data covered by the DPDP framework?
Recruitment involves digital personal data and the DPDP framework will become highly relevant, but many substantive provisions remain subject to phased commencement as of October 2026.
The Most Important Distinction: Automating a Process vs Automating Discrimination
AI can make recruitment faster.
A company may receive:
100,000 applications.
Human review of every CV may be impractical.
Automation can legitimately help:
- verify qualifications;
- identify relevant skills;
- organise applicants;
- schedule interviews.
But efficiency does not excuse illegality.
The employer cannot say:
“Our discriminatory screening would have been illegal if a human did it, but it is lawful because software did it.”
Automation does not change the substance of the decision.
The Second Important Distinction: Prediction vs Proof
AI hiring systems frequently predict:
“This candidate is likely to succeed.”
But prediction is not fact.
A model score is based on:
- assumptions;
- data;
- statistical correlations.
It should not automatically be treated as objective truth.
The more serious the employment consequence, the stronger the case for:
- validation;
- human review;
- transparency;
- appeal.
Conclusion
Can companies reject candidates using algorithms in India?
Yes, but AI does not create a legal safe harbour for employers.
India currently permits AI-assisted recruitment and has not enacted a blanket prohibition on automated candidate screening.
At the same time, AI hiring operates within an increasingly important legal and governance framework.
For public-sector recruitment, Articles 14 and 16 of the Constitution require:
- fairness;
- equality;
- non-arbitrariness;
- equal opportunity.
For private employers, applicable statutory protections include areas such as:
- gender discrimination under the Code on Wages;
- transgender discrimination;
- HIV-related discrimination;
- disability-related employment obligations.
India’s 2025–26 AI Governance Framework adds an important policy direction.
The Government has explicitly recognised risks involving:
- bias;
- discrimination;
- exclusion;
- opacity;
- unfair outcomes
and promotes responsible AI based on fairness, accountability, understandability and human-centred governance.
The DPDP framework will also become increasingly important for the enormous volume of personal data processed during recruitment.
However, the timeline matters.
Under the November 2025 commencement notification, many substantive DPDP provisions—including Sections 3–17—are subject to an 18-month transition period and are therefore not yet fully operative in October 2026.
For employers, the safest approach is not to ask:
“Can AI legally reject candidates?”
but:
“Can we explain and defend the criteria by which this candidate was rejected?”
A responsible recruitment system should be:
- relevant to the job;
- tested for bias;
- tested for accuracy;
- appropriately transparent;
- privacy-conscious;
- accessible;
- subject to meaningful human oversight.
For candidates, the central principle is equally simple:
A company may automate recruitment. It cannot automate unlawful discrimination and then blame the machine.
This article reflects the Constitution of India, Code on Wages, 2019, Transgender Persons (Protection of Rights) Act, 2019, Rights of Persons with Disabilities Act, 2016, HIV and AIDS (Prevention and Control) Act, 2017, Digital Personal Data Protection framework, India AI Governance Guidelines and publicly available legal and policy developments up to October 2026. India does not currently have a single dedicated statute governing every use of AI in recruitment, and liability depends on whether the employer is public or private, the characteristics affected, the data used and the role played by automation. This article is intended for general legal information and academic discussion and does not constitute legal advice.
Primary SEO keyword: AI Hiring in India
Secondary SEO keywords: AI recruitment India law, algorithmic hiring India, AI discrimination employment India, automated hiring India, AI interview legal India, AI recruitment bias India, algorithm reject candidate India, AI Governance Guidelines recruitment, DPDP Act recruitment data, AI hiring discrimination India 2026, automated job rejection India, public employment AI India.
