1. Introduction

Artificial Intelligence Genomics Governance refers to the legal, ethical, scientific and institutional framework governing the use of artificial intelligence (AI) in genomic research, genetic testing, genome sequencing, disease-risk prediction, gene editing and precision medicine.

AI genomics combines computational systems with information about an individual's DNA to identify genetic variations, predict disease susceptibility, discover potential drug targets and support medical decisions. Although these applications can improve diagnosis and treatment, they also raise serious questions concerning privacy, informed consent, discrimination, scientific reliability, ownership of genetic data, accountability for algorithmic errors and the regulation of genetic interventions.

The central legal question is:

Who should be legally responsible when an AI system interprets genomic information incorrectly, exposes confidential genetic data, or contributes to a harmful medical or genetic intervention?

There is no single, universally established legal doctrine called Artificial Intelligence Genomics Governance. Rather, the subject is governed by the interaction of existing laws on data protection, medical negligence, biotechnology, intellectual property, human rights and administrative regulation.

This explanation focuses on India, with comparative references to the United States, the United Kingdom and Europe. The cases discussed below establish relevant legal principles; they should not be mistaken for judgments that directly decided every issue involving AI and genomics.

2. Principal objectives of AI genomics governance

The framework has several interconnected objectives.

Genomic privacy: Prevent unauthorized collection, processing, disclosure or commercial exploitation of DNA-related information.

Scientific validity: Ensure that AI predictions are sufficiently accurate, validated and suitable for their intended clinical use.

Informed consent: Require meaningful consent for genomic testing, secondary research, data sharing and, where applicable, genetic interventions.

Non-discrimination: Protect individuals from unfair treatment based on genetic characteristics or predictions.

Professional accountability: Identify the responsibilities of clinicians, laboratories, researchers, software developers, hospitals and data controllers.

Biotechnology safety: Ensure that AI-supported genetic engineering and gene-editing activities comply with applicable biosafety and research regulations.

Equitable access: Prevent genomic medicine from systematically excluding populations underrepresented in genetic datasets.

These objectives must be balanced against legitimate scientific research, public health interests and the potential benefits of personalized medicine.

3. Legal framework in India

A. Constitutional protection of privacy and autonomy

The Indian Constitution provides an important foundation for regulating AI-based genomic systems.

Article 21: Protects life and personal liberty, including the constitutionally recognized right to privacy.

Article 14: Requires equality before the law and protection against arbitrary state action.

Article 19(1)(g): Protects lawful professional and commercial activity, subject to constitutional restrictions.

Article 32 and Article 226: Provide constitutional remedies through the Supreme Court and High Courts, respectively.

These provisions become relevant when government hospitals, public research institutions or state agencies use genomic AI in ways that threaten privacy, bodily autonomy, equality or procedural fairness.

B. Digital Personal Data Protection Act, 2023

The Digital Personal Data Protection Act, 2023 (DPDP Act), together with the Digital Personal Data Protection Rules, 2025, is an important part of India's developing framework for digital genomic information. The Rules were notified in November 2025, with implementation occurring in phases.

MeitY

+1

Relevant legal issues include:

Lawful processing: Genomic information in digital form must be handled according to applicable statutory requirements.

Consent and notice: Where consent is the applicable legal basis, individuals should receive the required information and provide valid consent.

Security safeguards: Organizations processing genetic information must address unauthorized access, disclosure and data breaches as required by applicable provisions.

Purpose limitation in practice: Research institutions should define their purposes and avoid using data beyond the lawful scope of their processing.

Rights and remedies: Applicable rights, complaint mechanisms and obligations depend on the provisions in force and the circumstances of processing.

A critical qualification is that the DPDP Act does not establish a separate, comprehensive category of genetic data equivalent to the former sensitive-personal-data classifications under earlier Indian rules. Nor does the notification of the Act mean that every substantive provision and rule became enforceable immediately. The applicable commencement date must be checked for the particular obligation.

MeitY

+2

C. Information Technology Act, 2000

Section 43A and Section 72A have historically been relevant to questions of inadequate security practices and unlawful disclosure of information. However, their present application must be assessed alongside the DPDP Act's commencement and repeal or transition provisions. They should not automatically be treated as the current, exclusive framework for every genomic-data incident.

D. Medical and biotechnology regulation

AI genomic applications may also engage:

The Drugs and Cosmetics Act, 1940, and applicable rules governing relevant medical products and testing.

The New Drugs and Clinical Trials Rules, 2019, where applicable.

The Indian Council of Medical Research's ethical guidance for biomedical and health research.

The Department of Biotechnology's applicable biosafety framework and rules governing genetically engineered organisms.

The applicable medical-professional standards, laboratory requirements and research ethics approvals.

A genomic prediction system used solely for exploratory research may raise different regulatory issues from a system used to guide diagnosis, prescribe treatment or select an embryo for a medical procedure. The legal classification depends on the product, intended purpose and activity involved.

4. Important case laws

The following decisions provide legal principles relevant to genomic AI governance. Most predate modern AI genomics and therefore operate by analogy rather than directly deciding liability for a genomic algorithm.

1. Justice K.S. Puttaswamy (Retd.) v. Union of India (2017)

Citation: (2017) 10 SCC 1 — Supreme Court of India.

Facts and issue: The Supreme Court considered whether privacy is protected as a fundamental right under the Indian Constitution.

Judgment: The Court recognized privacy as a constitutionally protected fundamental right, grounded in liberty, dignity and autonomy.

Relevance to AI genomics:

A person's genome can reveal information not only about that person but also about biological relatives, ancestry and potential health risks. Collecting or analyzing genomic information through AI can therefore affect multiple dimensions of privacy.

The decision supports scrutiny of genomic surveillance, compulsory genetic profiling and state access to genomic databases. State interference must satisfy the applicable constitutional requirements, including legality, legitimate purpose and proportionality.

Legal principle: Genomic data governance must respect constitutional privacy and individual autonomy, particularly where government action is involved.

2. Suchita Srivastava v. Chandigarh Administration (2009)

Citation: (2009) 9 SCC 1 — Supreme Court of India.

Facts and issue: The case concerned reproductive autonomy and whether a woman could be compelled to undergo an abortion.

Judgment: The Court recognized reproductive choice as an aspect of personal liberty under Article 21.

Relevance to AI genomics:

AI systems may predict genetic disorders in embryos or fetuses, estimate inherited disease risks or recommend reproductive interventions. Such predictions must not be treated as automatically determining a person's reproductive choices.

The case supports the importance of bodily autonomy and voluntary reproductive decision-making. It does not, however, decide the legality of embryo selection by AI or establish an unrestricted right to use any reproductive technology.

Legal principle: Algorithmic recommendations cannot replace legally protected human autonomy.

3. Samira Kohli v. Dr. Prabha Manchanda (2008)

Citation: (2008) 2 SCC 1 — Supreme Court of India.

Facts and issue: The Court considered the requirements of valid consent for medical treatment and procedures.

Judgment: The Court emphasized that consent must be real and valid, based on adequate information, and that consent for one procedure does not ordinarily authorize a substantially different procedure.

Relevance to AI genomics:

A patient who consents to genetic testing may not necessarily have consented to every secondary use of the resulting data. For example, consent for diagnosis does not automatically resolve whether an institution may use the same data to train a commercial AI model, share it with unrelated researchers or develop a separate predictive product.

The exact legal requirements depend on the applicable statute, research rules and the scope of the original consent.

Legal principle: Consent to a medical intervention should not be treated as blanket authorization for unrelated genomic research or data exploitation.

4. Association for Molecular Pathology v. Myriad Genetics, Inc. (2013)

Citation: 569 U.S. 576 (2013) — Supreme Court of the United States.

Facts and issue: The case concerned patents over isolated human genes, including BRCA1 and BRCA2, which are associated with increased risks of breast and ovarian cancer.

Judgment: The Court held that naturally occurring DNA does not become patent-eligible merely because it has been isolated. It distinguished naturally occurring DNA from synthetically created complementary DNA (cDNA), subject to the limits discussed in the judgment.

LII / Legal Information Institute

+1

Relevance to AI genomics:

The decision is relevant when a company claims intellectual-property rights over naturally occurring genetic sequences discovered or analyzed using AI.

However, it does not mean that all genomic AI inventions are unpatentable. Novel methods, technical applications, engineered materials and other inventions require separate analysis under the relevant patent law.

In India, patentability must be assessed under the Patents Act, 1970, including Section 3(c), which excludes the mere discovery of a scientific principle or naturally occurring living or non-living substance, and other applicable exclusions.

Legal principle: Computational discovery does not automatically confer exclusive ownership over naturally occurring genetic information.

5. Moore v. Regents of the University of California (1990)

Citation: 51 Cal. 3d 120 (California Supreme Court).

Facts and issue: The case concerned the use of biological materials obtained from a patient for medical research and the development of a commercially valuable cell line.

Judgment: The court rejected the patient's conversion claim over the excised cells under the circumstances presented, while allowing informed-consent and fiduciary-duty claims to proceed.

Relevance to AI genomics:

The case illustrates why ownership of biological material, rights over genetic information, consent to research and duties of disclosure must be analyzed separately.

For example, a hospital's lawful possession of a tissue sample does not by itself answer every question about whether it may use the sample to train an AI model, disclose identifiable genomic information or commercialize a resulting invention.

Legal principle: Ownership of a biological sample does not resolve all questions of consent, disclosure or commercial research obligations.

6. Havasupai Tribe v. Arizona Board of Regents (2008 settlement)

Nature: United States litigation resolved through a settlement rather than a final appellate precedent establishing a general rule of law.

Background: Members of the Havasupai Tribe challenged the secondary use of blood samples originally collected for research concerning diabetes. The dispute involved alleged use of samples for additional studies, including research concerning ancestry and other sensitive matters.

Relevance to AI genomics:

The dispute highlights the importance of research consent that adequately explains foreseeable secondary uses, especially when research involves Indigenous communities and collective cultural interests.

AI systems can generate additional inferences from existing genomic datasets that were not anticipated when the samples were first collected.

Legal principle: Researchers should not assume that collecting a sample for one stated purpose authorizes every later use. The enforceability of consent conditions depends on the relevant law and facts.

7. Sorrell v. IMS Health Inc. (2011)

Citation: 564 U.S. 552 (2011) — Supreme Court of the United States.

The case concerned state restrictions on the sale, disclosure and use of prescriber-identifying information for pharmaceutical marketing. The Court invalidated the challenged restrictions under the First Amendment.

Relevance: It illustrates the constitutional tensions that can arise when governments regulate the commercial use of health-related data. It does not establish a general right to commercialize genomic information or eliminate privacy protections.

In genomic AI disputes, the case is useful as a comparative reference when evaluating the interaction between data regulation, commercial interests and freedom of expression.

8. Gonzales v. Carhart (2007)

Citation: 550 U.S. 124 (2007) — Supreme Court of the United States.

The Court upheld a federal statutory restriction on a particular abortion procedure. The judgment addressed legislative regulation of medical procedures and the role of medical uncertainty.

Relevance: Its application to AI genomics is limited. It may be discussed in broader debates about legislative authority over emerging reproductive technologies, but it is not a direct authority on genetic testing, embryo selection or AI-based reproductive medicine. Current reproductive-rights law must also be considered.

9. National Federation of the Blind v. United States Department of Transportation is not a direct genomic AI precedent

For academic accuracy, it is important not to attach a case to a legal principle merely because it involves technology. AI genomic governance should rely on decisions that genuinely concern privacy, consent, discrimination, scientific evidence, patents or medical responsibility.

A more directly relevant comparative statute is the Genetic Information Nondiscrimination Act of 2008 (GINA) in the United States. It restricts specified uses of genetic information in health insurance and employment, subject to statutory exceptions and limitations. It does not comprehensively prohibit every form of genetic discrimination, and its protections do not extend identically to life, disability or long-term-care insurance.

5. Major legal issues in AI genomics governance

A. Genetic privacy and re-identification

Genomic information presents distinctive privacy risks. Removing a person's name from a dataset does not necessarily make the information impossible to link back to that individual, particularly when genomic records are combined with family trees, public databases or other identifying information.

Legal questions include:

Whether genomic data was collected and processed lawfully.

Whether the individual received adequate notice about research and commercial uses.

Whether a research institution has appropriate access controls and retention practices.

Whether a secondary use is permitted by the original consent and applicable law.

Whether a disclosure or data breach triggers notification, compensation or other remedies.

Example: A hospital uses de-identified DNA data to train a cancer-risk model. An external researcher later combines the data with genealogy records and identifies a patient. The legality of the original processing, the adequacy of safeguards, the researcher's conduct and any resulting harm must be examined separately.

B. Algorithmic bias and genetic discrimination

AI models trained on genomic datasets may perform less accurately for populations that were underrepresented in their training data.

This can create unequal access to accurate diagnosis, unnecessary medical interventions or incorrect disease-risk predictions.

A responsible governance framework should require, where appropriate:

Validation across relevant populations.

Documentation of the training data and its limitations.

Independent evaluation of clinically important errors.

Procedures for challenging consequential decisions.

Monitoring for performance deterioration and unequal outcomes.

In India, Article 14 may be relevant to discriminatory state action. Private-sector claims will depend on the applicable statutory protections and other legal causes of action. Genetic discrimination is not automatically unconstitutional in every private transaction.

C. Medical negligence and AI-generated errors

Suppose an AI system incorrectly classifies a genetic variant as pathogenic, leading a clinician to recommend unnecessary surgery.

Potentially responsible parties may include the clinician, hospital, laboratory, manufacturer or software provider. Liability is not automatic: it depends on the relevant duty, breach, causation, damage and any applicable statutory or contractual rules.

The investigation should consider:

Whether the system was appropriately validated.

Whether the clinician reviewed the output against clinical evidence.

Whether the system communicated uncertainty adequately.

Whether the patient received appropriate information.

Whether the error was foreseeable and avoidable.

Whether the harm would have occurred without the alleged breach.

AI output should not be treated as conclusive proof of a genetic diagnosis. Equally, clinicians should not be assumed negligent merely because an appropriately used system produces an unexpected result.

D. Intellectual property and genomic inventions

AI can identify genetic associations, propose molecular structures and help design gene-editing methods. Patentability must nevertheless be determined under the law of the relevant jurisdiction.

In India, important questions arise under the Patents Act, 1970, including:

Whether the claimed subject matter constitutes a patentable invention.

Whether it is novel and involves an inventive step.

Whether the invention is sufficiently disclosed.

Whether an applicable statutory exclusion applies.

Whether the invention involves the mere discovery of naturally occurring biological material.

The Myriad Genetics decision is a useful comparative authority, but Indian patentability must ultimately be assessed under Indian law rather than by mechanically applying US doctrine.

E. AI-assisted gene editing and biosafety

AI may help identify targets for CRISPR-based editing or predict the effects of genetic modifications. This creates legal questions about off-target changes, ecological consequences, inherited genetic alterations and responsibility for experimental harm.

In India, the applicable regulatory pathway depends on the organism, technology and proposed activity. Relevant instruments may include the Environment (Protection) Act, 1986, the rules governing hazardous microorganisms and genetically engineered organisms, and applicable institutional biosafety requirements.

A computational prediction does not itself authorize a genetic intervention. Laboratory, clinical, environmental and other approvals may be required independently.

6. Institutional responsibilities

The allocation of responsibility should reflect actual control over the activity, the duties imposed by law and each party's contribution to the harm. A contractual clause stating that a clinician or user accepts all risk does not necessarily eliminate statutory or professional responsibility.

7. International governance standards

International standards provide useful guidance for national regulators, researchers and healthcare organizations.

World Health Organization (WHO)

The WHO's Human Genome Editing: A Framework for Governance and accompanying recommendations, published in 2021, address oversight, international cooperation, research registries, unsafe or unregistered activities, public engagement and intellectual property.

World Health Organization

+1

These are governance recommendations, not a directly enforceable global statute.

European Union

The General Data Protection Regulation (GDPR) treats genetic data as a special category of personal data under Article 9. Processing generally requires an applicable exception to the prohibition, together with compliance with other GDPR requirements. Medical and scientific research provisions may be relevant, subject to their conditions and safeguards.

United States

GINA provides specific protections against genetic discrimination in employment and health insurance. It is an important comparative model, but it does not cover every insurance product or every form of genetic discrimination.

U.S. Equal Employment Opportunity Commission

+1

8. Practical example: AI genomic risk prediction

Consider a biotechnology company that develops an AI system predicting whether a patient carries a genetic variant associated with a serious inherited disease.

Data collection: The company obtains genomic samples and related clinical records. It must establish the lawful basis for collection and processing, appropriate consent where required, and compliance with applicable research and data-protection requirements.

Model development: Researchers train the model on genomic datasets. They should evaluate representativeness, accuracy, false-positive rates and false-negative rates.

Clinical deployment: Before the output is used in patient care, the organization must determine the applicable regulatory requirements and establish appropriate clinical oversight.

Harmful prediction: The AI incorrectly predicts a high disease risk, and the patient undergoes an unnecessary procedure. The parties' professional duties, system performance, causation, informed consent and the actual harm must be investigated.

Remedy: Depending on the facts, the patient may pursue a professional-negligence claim, a consumer-law remedy, a contractual claim, a data-protection complaint or another applicable legal remedy.

This example illustrates why governance must operate throughout the AI lifecycle rather than only when a product first enters the market.

9. Recommended legal safeguards

A comprehensive AI genomics governance framework should include the following measures:

Mandatory risk assessment: Require documented assessment before high-impact genomic AI is deployed.

Independent scientific validation: Evaluate clinical accuracy and limitations before consequential use.

Meaningful consent: Clearly distinguish clinical testing, secondary research, model training and commercial use.

Genomic-data security: Apply access controls, encryption, retention limits and appropriate breach-response procedures.

Human oversight: Require qualified professionals to review high-stakes results and permit appropriate challenges.

Fairness audits: Examine differences in model performance across relevant populations.

Traceability: Maintain records of model versions, input data, validation and consequential decisions.

Clear liability allocation: Define responsibilities among developers, laboratories, hospitals and researchers without improperly excluding mandatory legal duties.

Special oversight for heritable editing: Apply stringent ethical and regulatory review to research or interventions that may affect future generations.

These safeguards are policy recommendations; whether any particular measure is legally mandatory depends on the applicable jurisdiction, product classification and activity.

10. Conclusion

Artificial Intelligence Genomics Governance is an emerging field at the intersection of biotechnology, constitutional law, privacy, medical negligence, intellectual property and AI regulation.

In India, Puttaswamy provides a foundation for genomic privacy, Suchita Srivastava supports reproductive autonomy, and Samira Kohli establishes important principles of informed medical consent. The US decisions in Myriad Genetics and Moore offer comparative guidance on genetic patents and the legal distinction between biological materials, research consent and commercial interests.

The WHO's genome-editing governance framework further demonstrates the need for institutional oversight and international cooperation.

The fundamental legal principle is that technological capability does not itself create legal authority. An AI system's ability to analyze, predict or modify genetic information does not remove the need for lawful processing, scientific validation, informed decision-making, appropriate oversight and accountability for harm.

Academic note: The cases above are relevant authorities, but none should be cited as directly establishing a comprehensive legal regime for AI genomics. For litigation or formal academic submission, verify the current statutory text, commencement notifications and full judgments in the relevant jurisdiction.