AI-Generated Defamation Claims in India

1. Introduction

AI-generated defamation claims arise when an artificial-intelligence system generates, reproduces, transforms, or disseminates a false statement that harms the reputation of an identifiable person, company, organisation, or other legally protected claimant.

Indian law does not presently recognise a separate statutory tort called “AI-generated defamation.” The ordinary principles of defamation apply to content produced or disseminated through AI, together with additional questions concerning:

who is legally responsible for the AI output;

whether the statement is defamatory;

whether it identifies the claimant;

whether it was communicated to a third party;

whether the statement is false;

whether the defendant knew or should have known of the risk;

whether the AI output was merely generated privately or publicly disseminated;

whether intermediary/platform protections apply;

whether the claimant is a public figure;

whether the statement is protected opinion or a statement of fact.

AI therefore creates new factual and evidentiary problems, but the underlying legal principles of defamation remain largely familiar.

2. What Is an AI-Generated Defamatory Statement?

An AI-generated statement may be defamatory if it falsely attributes to a person conduct or characteristics that lower that person's reputation in the estimation of others.

Examples include an AI system falsely stating:

“X was convicted of financial fraud.”

when X was never convicted.

Or:

“Dr. A lost his medical licence because of professional misconduct.”

when that never happened.

Or an AI-generated article falsely stating:

“Company B falsified its accounts.”

These statements can potentially create civil and, depending on the applicable criminal law and facts, criminal defamation issues.

3. AI Does Not Automatically Become the Defendant

One of the first legal questions is:

Who is responsible for the defamatory output?

An AI system is not ordinarily treated as an independent legal person capable of bearing ordinary civil or criminal responsibility merely because it generated the words.

Potentially relevant actors include:

AI developer;

AI company;

website or platform operator;

publisher;

journalist using AI;

employer;

person who deliberately prompts the AI;

person who republishes the output;

human editor;

social-media account holder.

Liability therefore depends upon the role and conduct of the human or legal entity behind the publication.

4. Indian Legal Framework

AI-generated defamation can engage several areas of Indian law.

A. Civil defamation

Indian civil defamation is principally governed by common-law principles developed through judicial decisions.

Possible remedies include:

damages;

injunction;

removal of defamatory material;

correction/retraction;

declaration;

other equitable relief.

B. Criminal defamation

The criminal law framework has changed with the enactment of the Bharatiya Nyaya Sanhita, 2023 (BNS), which replaced the Indian Penal Code from 1 July 2024.

The offence of defamation is now dealt with under Section 356 BNS.

C. Constitutional law

Defamation claims must be balanced against:

Article 19(1)(a) — freedom of speech and expression;

Article 19(2) — reasonable restrictions, including defamation;

Article 21 — protection of reputation and dignity.

D. Information-technology law

Where AI-generated content is disseminated online, questions concerning intermediaries, electronic records, platform obligations and removal of unlawful content may arise.

5. Essential Elements of Defamation

A conventional defamation claim generally requires analysis of the following elements:

1. Defamatory imputation

There must be a statement capable of harming reputation.

2. Reference to the claimant

The statement must concern the claimant.

3. Publication

The statement must be communicated to someone other than the claimant in a conventional defamation action.

4. Falsity

Truth is generally a fundamental defence, subject to the applicable legal framework.

5. Absence of applicable privilege or other defence

Statements may be protected by:

absolute privilege;

qualified privilege;

fair comment/opinion;

truth;

statutory exceptions;

other recognised defences.

AI does not eliminate these requirements.

6. AI Hallucination and Defamation

One of the most important AI-specific problems is hallucination.

An AI system may confidently produce a completely fictional statement about a real person.

For example:

“Mr. X was arrested in 2023 for money laundering.”

The AI may have fabricated the allegation.

If the statement is shown to another person, the legal issue becomes significantly more serious.

The central questions become:

Who caused the publication?

Was the statement generated automatically?

Was it publicly displayed?

Did the provider know that hallucinations could occur?

Did it have safeguards?

Was there a reasonable opportunity for correction?

Did a human publish or republish the output?

7. Private AI Conversation vs Public Publication

This distinction is extremely important.

Situation A — Private interaction

A user asks an AI chatbot:

“Tell me about X.”

The chatbot privately responds with a false defamatory allegation.

The question of publication to a third party becomes complicated if only the user received the statement.

Traditional defamation requires communication to someone other than the claimant.

Situation B — Public AI-generated article

An AI system generates an article accusing X of fraud, and the article is published on a public website.

This creates a much stronger defamation case.

Situation C — AI-generated social-media post

A person prompts an AI to create a defamatory post and then publishes it.

The human publisher may face direct responsibility.

8. AI-Assisted Defamation

Not every case will involve completely autonomous AI.

There are several degrees of human involvement.

Level 1 — AI generates and publishes automatically

Potential issues concern the platform's role, system design and legal duties.

Level 2 — AI generates, human reviews and publishes

The human publisher's responsibility becomes significantly more important.

Level 3 — Human intentionally prompts AI to produce defamatory content

The human's conduct may be even more directly relevant.

Level 4 — Human copies AI's false statement and publishes it elsewhere

The republisher may face liability independently.

Thus, “the AI generated it” is not necessarily a complete defence.

9. Supreme Court Case: Subramanian Swamy v. Union of India

Subramanian Swamy v. Union of India, (2016) 7 SCC 221

This is one of the most important Indian authorities.

The Supreme Court upheld the constitutional validity of criminal defamation and recognised reputation as an important aspect of Article 21.

The Court also considered the relationship between:

freedom of speech under Article 19(1)(a);

defamation;

protection of reputation;

Article 19(2).

Relevance to AI

The principle is directly important for AI-generated defamatory content.

An AI system cannot be treated as making reputation legally irrelevant.

If an AI-generated statement falsely damages a person's reputation, constitutional protection of reputation remains relevant.

10. Case: M. C. Verghese v. T. J. Poonan

M.C. Verghese v. T.J. Poonan, AIR 1970 SC 1876

The Supreme Court considered principles concerning criminal defamation and publication.

The case is relevant to understanding the requirement of communication/publication and the legal consequences attached to defamatory statements.

AI relevance

The same conceptual question arises when AI produces content:

Who communicated the defamatory imputation, and to whom?

An AI system may generate text internally without meaningful third-party publication. Conversely, a system incorporated into a public platform may automatically disseminate the content.

11. Case: Jeffrey J. Diermeier v. State of West Bengal

Jeffrey J. Diermeier v. State of West Bengal, (2010) 6 SCC 555

The Supreme Court examined criminal defamation and the requirements associated with defamatory imputations and statutory exceptions.

The case is useful for analysing the distinction between:

mere criticism;

defamatory imputation;

protected statements;

statements falling within recognised exceptions.

AI relevance

AI systems frequently produce:

summaries;

criticism;

opinions;

rankings;

evaluations.

Not every negative AI-generated statement is defamatory.

The legal question remains whether the content constitutes a legally actionable defamatory imputation.

12. Case: S. Khushboo v. Kanniammal

S. Khushboo v. Kanniammal, (2010) 5 SCC 600

The Supreme Court emphasised that criminal law relating to reputation cannot be invoked merely because someone is offended by a statement.

The context and nature of the statement matter.

AI relevance

AI-generated content can be provocative, controversial or offensive without necessarily constituting defamation.

For example:

“X is a terrible actor.”

may be opinion.

But:

“X stole ₹10 crore from his employer.”

is a factual allegation capable of verification and potentially defamatory if false.

13. Case: R. Rajagopal v. State of Tamil Nadu

R. Rajagopal v. State of Tamil Nadu, (1994) 6 SCC 632

The Supreme Court considered privacy, publication and freedom of the press.

The case is highly relevant to modern AI because it illustrates the tension between:

freedom of publication;

individual privacy;

reputation;

public information.

AI relevance

An AI system may compile publicly available information about an individual and generate a narrative that contains defamatory or privacy-intrusive material.

The fact that individual pieces of information are publicly accessible does not automatically resolve every legal question concerning the resulting publication.

14. Case: Ram Jethmalani v. Subramaniam Swamy

Ram Jethmalani v. Subramaniam Swamy, 2006 (4) SCC 200

The Delhi High Court considered civil defamation principles and the importance of reputation.

The case is useful in understanding civil liability for statements that harm reputation.

AI relevance

AI-generated publications can similarly give rise to civil claims where the claimant establishes that the publication caused actionable reputational harm.

15. Case: Khushwant Singh v. Maneka Gandhi

Khushwant Singh v. Maneka Gandhi, 2002 (1) SCC 50

This authority is frequently discussed in Indian defamation jurisprudence concerning injunctions and freedom of expression.

It illustrates the need for courts to carefully balance:

protection of reputation;

freedom of expression;

prior restraint;

public interest.

AI relevance

A claimant may seek an injunction preventing further dissemination of an AI-generated defamatory statement.

Courts may nevertheless be cautious about granting broad prior restraints because freedom of expression remains constitutionally protected.

16. Case: Sewakram Sobhani v. R.K. Karanjia

Sewakram Sobhani v. R.K. Karanjia, (1981) 3 SCC 208

The Supreme Court dealt with criminal defamation and the relationship between publication, public interest and the statutory exceptions.

The case demonstrates that the context and purpose of publication are important.

AI relevance

An AI-generated investigative article may contain allegations concerning a public official or company.

The legal analysis cannot stop at:

“The statement is negative.”

It must also consider whether a recognised defence applies.

17. At Least Six Important Case Laws — Consolidated Table

18. Defamation Through AI Chatbots

Consider this hypothetical:

A user asks:

“Who is businessman X?”

The chatbot answers:

“X was convicted of fraud and served three years in prison.”

Assume the statement is completely false.

Potential questions include:

A. Is the statement defamatory?

Potentially yes, because it alleges serious criminal conduct.

B. Is X identifiable?

Yes, if the chatbot clearly refers to X.

C. Was there publication?

The answer depends upon who received the information and the legal theory advanced.

D. Who is responsible?

Potential candidates may include:

AI provider;

platform;

deployer;

person who republishes the statement.

E. Was there fault?

This may depend upon:

foreseeability of hallucination;

system safeguards;

representations concerning accuracy;

notice;

corrective mechanisms.

19. AI-Generated Fake Criminal Accusations

Criminal accusations are especially serious.

Examples:

“X committed murder.”

“X was arrested for terrorism.”

“X embezzled company funds.”

“X is a convicted sexual offender.”

Such allegations can produce severe reputational consequences.

A claimant may seek:

injunction;

damages;

removal;

correction;

criminal proceedings where statutory requirements are satisfied.

The seriousness of the allegation may also influence the assessment of reputational harm.

20. Defamation and Deepfakes

AI-generated defamation is not limited to text.

AI can create:

fake photographs;

manipulated videos;

synthetic voices;

fake interviews;

fabricated documents;

fake social-media posts.

For example, AI could generate a video apparently showing a politician accepting a bribe.

The content may be false, but viewers may reasonably believe it authentic.

This can create overlapping issues concerning:

defamation;

impersonation;

privacy;

copyright;

passing off;

fraud;

election-related regulation;

intermediary/platform responsibility.

21. AI Voice Cloning

Suppose AI creates a fake recording in which a businessperson appears to say:

“I bribed the regulator.”

If distributed publicly, the reputational impact may be substantial.

The claimant would need to establish:

authenticity or falsity;

identification;

publication;

defamatory meaning;

responsibility of the publisher;

causation and damage where relevant.

Expert evidence concerning synthetic media may become crucial.

22. AI Defamation Against Companies

Companies and commercial organisations may also suffer reputational harm.

For example, AI may falsely state:

“Company X manufactures counterfeit medicines.”

A company may seek civil remedies depending upon the applicable law.

However, questions concerning the legal status of corporate reputation, commercial loss, malice and damages will depend upon the particular cause of action and facts.

23. Public Figures

AI-generated statements concerning:

politicians;

actors;

judges;

business leaders;

celebrities;

academics;

may receive heightened constitutional scrutiny because public discussion and criticism have greater protection.

But public figures do not lose their right to reputation.

The important distinction remains:

Protected criticism

“I believe Minister X's economic policy is disastrous.”

versus:

Potentially defamatory factual allegation

“Minister X secretly received ₹5 crore from Company Y.”

The second is a factual allegation capable of proof or disproof.

24. Opinion vs Fact

AI systems frequently generate evaluative statements.

Compare:

“X is an incompetent manager.”

with:

“X falsified the company's financial statements.”

The first may be opinion depending on context.

The second makes a factual allegation.

Courts will examine the meaning conveyed in context, not merely the grammatical form.

25. Truth as a Defence

Truth is one of the most important protections against defamation liability.

If an AI-generated statement accurately states an established fact, a claimant generally cannot establish defamation merely because the information is embarrassing or damaging.

However, the defendant may need to establish the relevant legal requirements associated with the defence being invoked.

An AI provider cannot simply say:

“The model found it somewhere online.”

The provenance and reliability of the information may become important.

26. AI and Defamatory Misinformation

A major problem arises where an AI model repeats information found on unreliable websites.

For example:

Website A publishes false information.

AI system ingests or retrieves it.

AI repeats the allegation.

User republishes the AI response.

Each stage raises a different legal question.

The mere existence of information online does not necessarily make it truthful.

27. Republishing AI-Generated Defamation

Suppose a chatbot generates:

“X committed fraud.”

A journalist copies the statement into an article without verification.

The journalist's conduct may be analysed separately from the AI provider's conduct.

The journalist cannot necessarily escape responsibility simply by saying:

“I did not write it; AI wrote it.”

Publication and republication are independent factual events requiring separate analysis.

28. Intermediary Liability

Online platforms may invoke intermediary protections depending upon:

their statutory status;

actual role;

knowledge;

control;

due diligence;

applicable provisions;

whether the platform merely hosts content or materially participates in its creation.

Shreya Singhal v. Union of India, (2015) 5 SCC 1

The Supreme Court examined intermediary liability under the Information Technology Act and struck down Section 66A.

The judgment is highly relevant to online speech and intermediary regulation.

For AI systems, however, the critical question may be:

Is the platform merely hosting third-party content, or is it itself generating and delivering the allegedly defamatory content?

That distinction can materially affect the analysis.

29. AI Developer vs AI Deployer

This distinction is particularly important.

Suppose a bank uses a third-party AI model.

The model provider supplies the underlying technology.

The bank:

integrates it;

supplies customer data;

creates prompts;

determines how outputs are displayed;

publishes recommendations.

If defamatory content is produced, the bank may have a different legal position from the underlying AI developer.

The court may therefore examine control over the relevant publication.

30. Notice and Failure to Correct

Suppose an AI platform is informed:

“Your chatbot is falsely stating that I was convicted of fraud.”

The provider then does nothing despite repeated notice.

The legal significance of subsequent conduct may differ from the initial generation of the statement.

Important questions include:

Was notice properly given?

Was the statement demonstrably false?

Could it be corrected?

Did the provider have practical control?

Did it continue to display the content?

Was the failure to correct unreasonable?

31. Repetition Risk

Generative AI can create a particularly serious problem: repeated hallucination.

If users repeatedly ask:

“What crimes did X commit?”

the system may repeatedly generate the same false answer.

This could produce multiple instances of dissemination.

From a liability perspective, the claimant may argue that once the provider became aware of the defect, continued generation was no longer an isolated error but an ongoing risk.

32. Evidentiary Problems

AI-defamation cases will require unusual evidence.

Important evidence may include:

complete AI conversation;

prompts;

system instructions where discoverable;

date and time;

model/version;

output;

screenshots;

URLs;

publication records;

metadata;

server logs;

correction requests;

notices sent to the provider;

responses from the provider;

expert reports.

A screenshot alone may not always establish the full technical context.

33. Expert Evidence

Technical experts may be required to determine:

whether content was AI-generated;

whether an image/video is synthetic;

whether audio was cloned;

whether the output could have been generated by the alleged model;

whether the system had known hallucination characteristics;

whether safeguards were technically feasible.

Expert evidence can therefore become central to establishing causation and attribution.

34. Burden of Proof

The burden will depend upon the precise claim.

In civil proceedings, the claimant ordinarily must establish the elements of the cause of action on the applicable standard.

In criminal defamation, the prosecution/complainant must satisfy the applicable criminal-law requirements.

AI-generated content creates additional factual disputes concerning:

authorship;

publication;

authenticity;

responsibility;

intent/knowledge where relevant.

35. Damages

Potential civil damages may take account of:

seriousness of allegation;

extent of publication;

duration;

prominence;

claimant's existing reputation;

actual financial loss;

professional consequences;

emotional/reputational injury;

conduct of defendant after notice.

AI-generated content can potentially reach thousands or millions of people rapidly, which may increase the practical significance of publication.

36. Injunctions

A claimant may seek an injunction to prevent continued publication.

Courts must balance:

Reputation

against

Freedom of expression.

This is particularly important where the allegedly defamatory content concerns:

political criticism;

journalism;

public-interest investigations;

artistic expression;

academic debate.

An injunction should not automatically become a mechanism for suppressing legitimate criticism.

37. Article 19(1)(a) and Article 19(2)

Indian constitutional law requires a balance.

Article 19(1)(a)

Protects freedom of speech and expression.

Article 19(2)

Permits reasonable restrictions, including in relation to defamation.

Therefore:

AI-generated content does not receive an automatic constitutional immunity.

But:

Protection of reputation does not mean that all AI-generated criticism can be prohibited.

The nature, truth, context and purpose of the statement remain crucial.

38. Article 21 and Reputation

The Supreme Court's decision in Subramanian Swamy v. Union of India is particularly important because it recognised reputation as an aspect of Article 21.

Thus, AI-related reputational harm can potentially implicate constitutional values even though the immediate dispute may be a private defamation claim.

39. Criminal Defamation Under the BNS

Following the transition from the IPC to the Bharatiya Nyaya Sanhita, 2023, criminal defamation is governed by Section 356 BNS.

Therefore, contemporary AI-generated criminal-defamation analysis should not simply cite the old IPC section as though it remains the operative provision.

The underlying concepts concerning:

imputation;

reputation;

publication;

intention/knowledge;

exceptions;

remain central to the analysis.

40. AI and Defamation of Deceased Persons

AI systems may generate false allegations concerning deceased individuals.

Defamation law contains specific statutory considerations concerning imputations concerning deceased persons, including where the imputation would have harmed the person's reputation if living and is intended to hurt specified interests.

Accordingly, AI-generated historical narratives may create legal issues even where the individual is no longer alive.

41. AI and Defamation of Groups

Statements about broad groups require careful analysis.

For example:

“Everyone in Profession X is corrupt.”

is different from:

“Mr. X, a member of Profession X, accepted a ₹10 lakh bribe.”

The second identifies an individual.

Group defamation involves additional questions concerning whether the class is sufficiently identifiable and whether the legal requirements for a particular claim are satisfied.

42. Defamation and Privacy Overlap

An AI system may produce a statement that is simultaneously:

false;

defamatory;

private;

highly sensitive.

For example:

“X secretly has a particular medical condition.”

Even if the statement were not defamatory in the conventional sense, publication could potentially raise privacy concerns.

Puttaswamy therefore provides an important constitutional background to AI-generated personal information.

43. AI Search Engines and Summarisation

AI search systems may summarise web pages and generate an answer such as:

“According to multiple reports, X committed fraud.”

The legal question becomes more complicated where the underlying reports themselves are inaccurate.

A summary is not necessarily insulated from liability simply because it attributes information to other sources.

The exact wording, attribution and context become important.

44. AI News Generation

AI-generated journalism creates several risks:

fabricated sources;

invented quotations;

incorrect attribution;

false allegations;

outdated information;

misleading summaries.

If a media organisation publishes AI-generated defamatory material, traditional principles of publisher responsibility may apply.

Human editorial supervision may therefore become a major factor.

45. Corporate Compliance and AI Defamation

Organisations using generative AI should establish:

verification requirements;

prohibited prompts;

fact-checking;

human review;

source verification;

publication controls;

correction mechanisms;

complaint handling;

logging;

AI-use disclosure policies.

These measures can reduce both the probability of defamatory output and litigation risk.

46. Possible Defences

A defendant may rely upon one or more of the following, depending on the circumstances:

Truth

The statement is substantially true.

Fair comment / opinion

The statement is protected opinion based on facts.

Privilege

The publication falls within an applicable privilege.

Public interest

The publication may be protected where the relevant legal requirements are satisfied.

No publication

The statement was not communicated to a third party in the legally relevant sense.

No identification

The statement does not reasonably identify the claimant.

No defamatory meaning

The words do not carry a defamatory meaning.

Lack of responsibility

The defendant did not author, publish or control the relevant content.

Intermediary protection

A platform may invoke statutory protection where the requirements are satisfied.

47. “The AI Did It” Is Not a Complete Defence

A central principle for future AI litigation is likely to be:

Delegating content generation to an AI system does not necessarily eliminate legal responsibility for the human or organisation that chose to deploy, publish or disseminate that content.

If a newspaper uses AI to write an article, it remains a newspaper's publication.

If a company deliberately asks AI to produce allegations about a competitor and publishes them, the company's conduct remains legally relevant.

If a platform automatically generates defamatory statements without human intervention, however, questions of foreseeability, control, statutory intermediary protection and system design become considerably more complicated.

48. Practical Legal Test for an AI-Generated Defamation Claim

A useful framework is:

Step 1 — Identify the exact AI output

Preserve the precise words, image, video or audio.

Step 2 — Identify the claimant

Determine whether a reasonable reader/viewer would understand the content to refer to the claimant.

Step 3 — Establish defamatory meaning

Does the content lower reputation?

Step 4 — Establish falsity

Can the alleged factual assertion be disproved?

Step 5 — Establish publication

Who received or accessed the material?

Step 6 — Identify the publisher

Was it:

AI provider;

user;

employer;

media organisation;

platform;

website?

Step 7 — Establish responsibility

Who controlled or authorised dissemination?

Step 8 — Examine defences

Consider:

truth;

opinion;

privilege;

public interest;

statutory exceptions.

Step 9 — Establish harm

Document:

professional consequences;

business losses;

social consequences;

publication reach.

Step 10 — Select remedy

Consider:

damages;

injunction;

correction;

removal;

criminal complaint where appropriate.

49. Hypothetical Example

Assume an AI-powered news platform publishes:

“Businessman A was arrested for laundering ₹50 crore through shell companies.”

The statement is completely false.

Businessman A discovers it and immediately notifies the platform.

Potential claim

A may argue:

Defamatory imputation: allegation of serious financial crime.

Identification: Businessman A is specifically named.

Publication: millions of users accessed the article.

Falsity: no arrest or prosecution occurred.

Responsibility: platform generated and published the article.

Damage: investors and business partners withdrew.

Notice: platform was informed but continued displaying it.

Potential remedies could include:

injunction;

removal;

correction;

damages;

other appropriate civil relief;

criminal proceedings if the requirements of Section 356 BNS and applicable procedural law are satisfied.

The platform could respond by arguing:

it was merely an intermediary;

the content was generated by a third-party user;

it lacked control;

statutory protection applies;

it promptly removed the material;

there was no requisite fault.

The outcome would depend heavily on the platform's actual technical and legal role.

50. Key Legal Challenges Created by AI

AI-generated defamation creates several novel questions that Indian courts will increasingly have to address:

1. Who is the publisher?

2. Is automated generation equivalent to human publication?

3. How should foreseeability be assessed?

4. Does repeated hallucination create continuing responsibility?

5. When does an AI provider assume responsibility for output?

6. How should platform/intermediary protections apply?

7. What evidence proves AI authorship?

8. How should synthetic audio/video be authenticated?

9. What damages should be awarded for viral AI-generated misinformation?

10. What degree of human supervision should reasonably be expected?

51. Important Distinction: AI Error vs AI Defamation

The distinction can be expressed simply:

AI error

“The AI incorrectly stated that Company X's revenue was ₹100 crore.”

This may be an accuracy problem.

AI defamation

“The directors of Company X stole ₹100 crore from the company.”

This is a potentially defamatory factual allegation because it attributes serious misconduct to identifiable persons.

Therefore:

Not every AI hallucination is defamation, but some AI hallucinations can constitute defamatory imputations when the legal elements of defamation are satisfied.

52. Conclusion

AI-generated defamation claims in India are presently governed primarily by existing defamation, constitutional, information-technology and procedural principles rather than by a separate AI-defamation statute.

The most important Supreme Court authority is Subramanian Swamy v. Union of India, (2016) 7 SCC 221, which confirms the constitutional significance of reputation and the validity of criminal defamation.

Other important authorities include:

M.C. Verghese v. T.J. Poonan, AIR 1970 SC 1876;

Jeffrey J. Diermeier v. State of West Bengal, (2010) 6 SCC 555;

S. Khushboo v. Kanniammal, (2010) 5 SCC 600;

R. Rajagopal v. State of Tamil Nadu, (1994) 6 SCC 632;

Ram Jethmalani v. Subramaniam Swamy, 2006 (4) SCC 200;

Khushwant Singh v. Maneka Gandhi, 2002 (1) SCC 50;

Sewakram Sobhani v. R.K. Karanjia, (1981) 3 SCC 208;

Shreya Singhal v. Union of India, (2015) 5 SCC 1.

The central legal principle is that AI-generated content does not exist outside ordinary legal responsibility. Courts are likely to focus on the human or corporate actors who designed, deployed, prompted, reviewed, published, controlled or failed to correct the allegedly defamatory material.

The most difficult future cases will be those involving autonomous AI publication, hallucinated criminal allegations, deepfake videos, AI-generated news, repeated chatbot defamation, intermediary liability and attribution of responsibility between the AI developer and the entity deploying the AI.