Civil Law and AI-Generated Fake News Economic Damage Litigation in Europe

1. Introduction

AI-generated fake news economic damage litigation concerns situations where artificial intelligence generates, alters, or amplifies false information and that information causes measurable economic harm to an individual, company, professional, investor, institution, or market participant.

Examples include:

AI generating a false report that a company has become insolvent;

an AI chatbot falsely stating that a business committed fraud;

a deepfake announcing that a CEO has resigned;

an AI-generated story falsely claiming that a bank is failing;

fabricated financial results being circulated online;

false AI-generated allegations causing customers to abandon a business;

AI-generated misinformation causing cancellation of contracts;

fake news causing a company's share price to fall;

AI-generated false product-safety claims;

false information damaging a professional's ability to obtain contracts;

automated recommendation systems massively amplifying fabricated information.

European law does not have one single tort called “AI fake-news liability.” The claim usually has to be constructed through existing doctrines such as:

defamation;

protection of personality/reputation;

unfair competition;

negligence;

contractual liability;

data protection;

consumer protection;

intermediary/platform liability;

intellectual-property law;

market-abuse rules where financial markets are affected.

The central legal question is:

Who is legally responsible when AI creates false information and that information causes economic damage?

2. The Basic Liability Chain

A typical dispute can be represented as:

AI system

↓

False statement generated

↓

Publication or dissemination

↓

Audience receives information

↓

Reliance / reputational effect / market reaction

↓

Economic damage

↓

Civil claim

The claimant normally has to establish the relevant elements required by the applicable national law.

A useful formula is:

AI Fake-News Liability

False or unlawfully harmful statement

Attributable publication/dissemination

Wrongfulness

Causation

Recognised economic or reputational damage

=

Potential civil liability

3. Important Preliminary Point: “Fake News” Is Not Automatically Unlawful

A false statement may be legally significant, but falsity alone does not automatically create liability in every European jurisdiction.

Courts commonly distinguish between:

Statements of fact

Example:

“Company X became insolvent on 1 September.”

This can normally be tested for truth or falsity.

Value judgments/opinions

Example:

“Company X is badly managed.”

This is generally not capable of being proved true or false in the same way.

The ECtHR repeatedly emphasises this distinction. For example, Sorguç v Turkey noted that factual assertions are capable of proof, whereas value judgments generally are not, although a value judgment may still require an adequate factual basis. (HUDOC)

This distinction becomes particularly important with generative AI because AI outputs frequently mix:

factual assertions;

predictions;

opinions;

summaries;

invented information.

4. Why AI Creates a Special Problem

Traditional false-information litigation usually involves:

Human author → publisher → audience

AI can create:

User prompt → AI model → generated statement → automated platform → recommender system → millions of recipients

There may therefore be several potentially responsible actors:

person who prompted the AI;

AI developer;

AI deployer;

publisher;

platform;

search engine;

recommender system;

advertiser;

person who deliberately amplified the false content.

Determining which actor legally caused the damage is therefore one of the most difficult issues.

5. European Legal Framework

A. Article 10 ECHR

Article 10 protects freedom of expression.

Therefore, courts must balance:

freedom of expression

against

reputation and other protected rights.

A defamation claim cannot automatically prevail merely because a statement is damaging.

B. Article 8 ECHR

Article 8 protects private life and reputation in appropriate circumstances.

This can become relevant when AI-generated misinformation seriously damages:

personal reputation;

professional identity;

private life;

social standing.

C. EU Charter

Relevant rights include:

Article 7 — private life;

Article 8 — personal data;

Article 11 — expression and information;

Article 16 — freedom to conduct a business;

Article 17 — property;

Article 47 — effective judicial protection.

6. Digital Services Act

The Digital Services Act is particularly important where fake news is distributed through very large online platforms or search engines.

The DSA expressly identifies systemic risks involving misleading or deceptive content, including disinformation, and requires very large platforms/search engines to consider algorithmic amplification, recommender systems, advertising systems, bots and coordinated manipulation in their systemic-risk assessments. (EUR-Lex)

The DSA also requires proportionate and effective risk-mitigation measures, while taking account of fundamental rights. (EUR-Lex)

Important distinction

However:

A DSA violation does not automatically establish a private damages claim for every person who suffers economic loss.

The claimant still needs an appropriate legal cause of action and must satisfy the applicable requirements for compensation.

7. EU AI Act

The AI Act adds another layer where AI systems are covered by its requirements.

For general-purpose AI models with systemic risk, the EU framework contains risk-assessment and mitigation obligations. The relationship between the AI Act and DSA is particularly important where an AI model is integrated into a very large online platform or search engine. The European Commission has specifically identified the interaction between AI-system systemic-risk obligations and DSA systemic-risk obligations. (EUR-Lex)

Thus, future litigation may involve:

AI Act + DSA + GDPR + national civil law + Article 8/10 ECHR

simultaneously.

8. Case 1 — Delfi AS v Estonia

ECtHR Grand Chamber, 16 June 2015

Application No. 64569/09

This is one of the most important European authorities concerning online publication and intermediary liability.

Delfi operated a commercial online news portal allowing readers to post comments. Some comments were seriously unlawful and offensive. The Estonian courts held Delfi liable.

The ECtHR found no violation of Article 10, emphasising factors including the commercial nature of the portal, the nature of the comments, the platform's control over the comments and the availability of measures for removing them. (HUDOC)

AI relevance

AI-generated fake news creates an analogous intermediary question:

When does a platform become legally responsible for harmful content generated or amplified through its system?

Delfi demonstrates that platform liability depends upon the role and circumstances of the intermediary, rather than a universal rule that platforms are always or never responsible.

Economic damage example

A platform's AI recommendation system repeatedly amplifies a false allegation:

“Company X is selling dangerous products.”

Customers cancel orders.

The company suffers €5 million in losses.

The company may attempt to establish:

falsity;

unlawful publication;

platform involvement;

causation;

economic damage.

9. Case 2 — Steel and Morris v United Kingdom

ECtHR, 15 February 2005

Application No. 68416/01

The case concerned defamation proceedings arising from allegations made against McDonald's.

The ECtHR examined the fairness and proportionality of the proceedings under Article 10.

One particularly important principle concerns the relationship between defamation damages and the injury suffered.

The Court stated that damages must bear a reasonable relationship of proportionality to the reputational injury. (HUDOC)

AI relevance

Suppose an AI-generated false story says:

“Company X committed fraud.”

The statement is reproduced 50 million times.

The claimant seeks enormous damages.

The court must consider:

seriousness of the allegation;

reach;

actual reputational harm;

economic consequences;

defendant's conduct;

proportionality of the award.

Important lesson

AI's enormous ability to generate and reproduce information makes quantification of damages particularly important.

10. Case 3 — Axel Springer AG v Germany

ECtHR Grand Chamber, 7 February 2012

This case concerned publication of information about a public figure and the balance between reputation/privacy and freedom of expression.

The ECtHR developed important balancing factors including:

contribution to a debate of general interest;

how well known the person was;

subject matter of the report;

prior conduct;

content and form;

consequences of publication;

severity of the sanction. (HUDOC)

AI relevance

Suppose an AI system generates:

“CEO X secretly manipulated company accounts.”

The statement concerns an issue of public and economic interest.

The court should not simply ask:

“Was it damaging?”

It must examine the competing rights and the circumstances surrounding publication.

For AI-generated journalism, therefore, public interest and responsible publication remain important.

11. Case 4 — UJ v Hungary

ECtHR, 19 July 2011

This case concerned criticism of a state-owned company.

The Court distinguished between:

factual allegations;

value judgments;

commercial reputation;

matters of public interest.

The Court noted that commercial reputation is different from the moral reputation of a private individual. (HUDOC)

AI relevance

This distinction is very useful in AI economic-damage cases.

A company may claim:

“AI-generated fake news destroyed our commercial reputation.”

The court may need to determine:

Is the claimant an individual or company?

Is the statement factual or opinion?

Does it concern a matter of public interest?

Is the alleged harm primarily economic?

Was the statement supported by any factual basis?

12. Case 5 — Timpul Info-Magazin and Anghel v Moldova

ECtHR, 27 November 2007

The case concerned allegedly defamatory allegations published by a newspaper.

The ECtHR found a violation of Article 10 and awarded compensation for material and moral damage. (HUDOC)

The case is useful for understanding the relationship between:

defamation;

publication;

freedom of expression;

compensation.

AI relevance

Imagine a generative-AI news service produces an invented allegation:

“Businessman X paid a €500,000 bribe.”

If the allegation is false and causes:

cancellation of contracts;

loss of customers;

financial losses;

the claimant may pursue remedies under applicable national defamation/civil-liability rules.

Timpul illustrates that courts must consider both the truthfulness/factual basis of the allegation and the proportionality of the interference with expression.

13. Case 6 — Kasabova v Bulgaria

ECtHR, 19 April 2011

The case concerned defamation proceedings against a journalist.

The Court examined:

factual allegations;

the possibility of proving truth;

procedural fairness;

severity of sanctions;

proportionality of damages.

The ECtHR reiterated that a defendant should have a realistic opportunity to establish the factual basis of allegations, and that damages must bear a reasonable relationship to the reputational injury. (HUDOC)

AI relevance

This is particularly important because AI-generated statements may be accompanied by:

fabricated “sources”

or

invented citations.

A claimant may argue that the AI output looks like a verified factual report even though the underlying facts never existed.

The legal analysis must therefore distinguish:

actual evidence

from

AI-generated appearance of evidence.

14. Case 7 — Google Spain v AEPD and Mario Costeja González

CJEU, Case C-131/12

This landmark case concerned Google's search engine and personal information.

The CJEU recognised that search engines play an important role in processing and disseminating personal information and established circumstances in which individuals could request removal of search results concerning personal data. (Infocuria)

AI relevance

Modern AI search systems do more than merely retrieve information.

They may:

summarise information;

combine sources;

generate answers;

rank sources;

create synthetic biographies;

reproduce allegations.

Suppose an AI search assistant repeatedly states:

“Person X was convicted of fraud.”

when no such conviction occurred.

Potential issues can include:

data accuracy;

privacy;

reputation;

correction;

de-indexing;

AI-generated misinformation.

Google Spain therefore provides an important foundation for analysing AI systems that organise and reproduce personal information.

15. Case 8 — L'Oréal v eBay

CJEU, Case C-324/09

Although primarily an intellectual-property case, the judgment is important for online-intermediary responsibility.

The CJEU examined the responsibility of an online marketplace and recognised circumstances in which courts could order measures against intermediaries to prevent continuing infringements. (Infocuria)

AI relevance

The case helps establish the broader proposition that an online intermediary's role matters.

An AI platform that merely provides technical infrastructure may be legally different from a platform that:

actively promotes content;

ranks it;

monetises it;

targets it;

amplifies it;

knows about its unlawful character.

Therefore:

The more active the platform's role, the more complicated intermediary-liability analysis becomes.

16. Case 9 — Sorguç v Turkey

ECtHR, 23 June 2009

The Court distinguished factual assertions from value judgments and emphasised the importance of factual foundation.

A value judgment can receive strong protection, but an opinion without an adequate factual basis may still create proportionality concerns. (HUDOC)

AI relevance

Generative AI frequently produces statements such as:

“Experts believe Company X is about to collapse.”

This appears to be an opinion.

But if no experts actually made the statement, the output may effectively contain a fabricated factual representation.

Thus the court must examine the substance of the AI output rather than simply its grammatical form.

17. Economic Damage Categories

AI-generated fake news can produce several categories of economic harm.

A. Lost sales

Customers stop buying products.

B. Lost contracts

Business partners terminate agreements.

C. Share-price impact

False financial information affects market valuation.

D. Increased financing costs

Banks or investors react to false information.

E. Loss of customers

Consumers believe fabricated allegations.

F. Reputational damage

The business loses commercial goodwill.

G. Professional income loss

A professional loses clients because of false allegations.

H. Emergency mitigation costs

The claimant spends money on:

public-relations responses;

legal notices;

crisis management;

cybersecurity;

corrective advertising.

18. Proving Economic Loss

This is one of the hardest parts of the litigation.

The claimant should ideally establish:

Step 1 — False statement

What exactly did the AI generate?

Step 2 — Publication

Where was it published?

Step 3 — Reach

How many people saw it?

Step 4 — Falsity

Why was it false?

Step 5 — Attribution

Who generated, published or amplified it?

Step 6 — Causation

Did the fake information cause the loss?

Step 7 — Quantification

How much economic loss resulted?

19. Causation Problem

Suppose:

AI generates false insolvency report.

Then:

Company's sales fall 20%.

But the company was already experiencing declining sales.

The claimant must distinguish:

AI-caused loss

from

pre-existing market decline.

Economic experts may therefore need to compare:

sales before publication;

sales after publication;

unaffected markets;

customer behaviour;

share-price movements;

competitor performance;

media coverage.

20. AI Amplification

AI creates an additional causation problem.

A false statement may begin with one user.

But an algorithm may then amplify it:

User posts fake story

↓

AI recommendation system identifies engagement

↓

Algorithm promotes story

↓

10,000 users see it

↓

100,000 users see it

↓

News aggregators reproduce it

↓

Financial consequences occur

The claimant may therefore argue that algorithmic amplification materially contributed to the damage.

The DSA expressly recognises that algorithmic amplification of misleading or deceptive content can contribute to systemic risks for very large platforms and search engines. (EUR-Lex)

21. AI Hallucination vs Deliberate Fake News

This distinction is essential.

AI hallucination

The system generates false information without the user necessarily intending deception.

Example:

AI invents a false court judgment.

Deliberate AI misinformation

A person intentionally prompts AI:

“Create a convincing fake report proving Company X is bankrupt.”

The resulting material is deliberately disseminated.

AI amplification

The original false material is created by a human, but AI recommendation systems dramatically increase its reach.

These three situations may lead to very different liability analyses.

22. Who May Be Liable?

A. Person who created the fake content

Potentially liable where national law establishes:

defamation;

intentional wrongdoing;

negligence;

unfair competition;

other civil wrongs.

B. AI developer

Developer liability is more complicated.

The claimant may need to establish a specific legal basis, such as:

defective product/service;

contractual breach;

negligence;

regulatory breach where it creates a relevant cause of action.

The mere fact that an AI model can hallucinate does not automatically make its developer liable for every false output.

C. Platform

The platform's liability depends on:

its legal role;

knowledge;

control;

applicable intermediary rules;

notice;

amplification;

moderation;

contractual relationship.

D. Publisher

If a media outlet republishes an AI-generated false story, traditional media-liability rules may apply.

23. AI Developer and Foreseeability

A future litigation question will be:

Was the economic harm reasonably foreseeable?

For example:

If a company deliberately markets an AI system as:

“A fully autonomous financial news generator”

but the system regularly invents financial events, claimants may argue that the risk of economic misinformation was foreseeable.

But foreseeability alone will not necessarily establish liability; the claimant still needs a legally recognised cause of action.

24. Fake Financial News

This is an especially serious category.

Examples:

“Bank X is insolvent.”

“Company Y has secretly lost €10 billion.”

“CEO Z has resigned.”

Such statements can cause:

bank runs;

share-price movements;

investor losses;

contract cancellations;

credit downgrades.

Where securities markets are involved, additional EU financial-market rules may become relevant, including prohibitions concerning market manipulation and dissemination of false or misleading information.

Civil litigation may therefore overlap with:

defamation + securities law + market abuse + regulatory enforcement.

25. AI Deepfakes

Deepfake video can create particularly severe economic harm.

Example:

A fabricated video shows a company's CEO saying:

“Our company has no money left.”

The video is fake.

Investors react.

Share price collapses.

Potential claims could concern:

defamation;

economic torts;

market manipulation;

fraud;

data/personality rights;

platform responsibility.

The evidentiary issue is easier if forensic evidence establishes that the video was synthetically generated.

26. AI Fabricated Documents

Generative AI can create fake:

court judgments;

regulatory decisions;

government announcements;

corporate filings;

financial reports;

contracts;

certificates.

This can produce economic harm even without a conventional defamatory statement.

For example:

AI generates a fake government notice stating that a factory has been closed.

Customers cancel orders.

The legal claim could involve misrepresentation, negligence, fraud or other national-law causes of action.

27. Economic Damage to Companies

Companies can have legally protected commercial interests.

The ECtHR has recognised that commercial reputation can be legally relevant while distinguishing it from the moral dimension of an individual's reputation. UJ v Hungary is useful on this distinction. (HUDOC)

A company may therefore claim damage involving:

goodwill;

customer relationships;

contracts;

commercial reputation;

market position.

But the precise availability of damages depends upon national law.

28. Economic Damage to Individuals

AI-generated fake news can damage:

employment;

professional licences;

business opportunities;

customer relationships;

investment opportunities;

creditworthiness.

Example:

AI falsely states that a lawyer was convicted of bribery.

Potential consequences:

clients leave;

law firm terminates relationship;

professional reputation falls;

income decreases.

This can create overlapping claims concerning:

reputation + privacy + economic loss.

29. Defamation and Freedom of Expression

Courts must balance:

Claimant

reputation;

privacy;

business interests;

economic rights.

Defendant

freedom of expression;

journalism;

public-interest reporting;

criticism.

The ECtHR's case law shows that the balance depends on factors such as:

public interest;

factual basis;

status of the person;

manner of publication;

seriousness of allegation;

consequences;

severity of sanction.

This balancing approach appears in cases such as Axel Springer, Kasabova, Sorguç, and Steel and Morris. (HUDOC)

30. AI and Duty of Care

A negligence-based claim may involve:

Duty of care

↓

Breach

↓

Causation

↓

Damage

Possible allegations could include:

inadequate verification;

failure to implement safeguards;

reckless deployment;

failure to correct known false information;

negligent amplification.

But whether a duty exists depends on the applicable national civil law.

31. AI Platform's Knowledge

Knowledge can be important.

There is a major difference between:

Situation A

Platform has no knowledge of the false statement.

Situation B

Platform receives a detailed notice proving the statement is false.

Situation C

Platform's AI system itself detects that the information is probably false.

Situation D

Platform knows the information is false but continues aggressively amplifying it.

The legal analysis may become increasingly demanding as the platform's knowledge and control increase.

32. Notice-and-Action

An injured company might notify the platform:

“This AI-generated story is false. Here is the evidence.”

The platform's response can become important evidence.

Possible outcomes:

immediate removal;

fact-check label;

reduced recommendation;

refusal to act;

continued amplification.

A documented failure to respond appropriately can become relevant to liability depending upon the applicable law.

33. Evidence in AI Fake-News Litigation

The claimant should preserve:

original AI output;

prompts;

timestamps;

URLs/platform records;

screenshots;

copies of generated videos;

metadata;

server logs;

reposting records;

engagement statistics;

platform recommendation data;

financial records;

customer cancellations;

investor communications;

expert forensic reports.

Critical point

AI content can disappear or change rapidly.

Therefore, evidence preservation is particularly important.

34. Expert Evidence

Economic-damage litigation may require experts in:

AI forensics

To establish:

whether content was AI-generated;

which system generated it;

whether the material was manipulated.

Digital forensics

To establish:

dissemination;

timestamps;

accounts;

reach.

Economics

To calculate:

lost sales;

market value;

lost contracts;

abnormal financial losses.

Media analytics

To measure:

audience reach;

algorithmic amplification;

engagement.

35. Damages

Possible damages may include:

Actual financial loss

For example:

lost sales;

cancelled contracts;

lost investment;

increased financing costs.

Loss of profit

Expected profit lost because of the misinformation.

Reputation-related loss

Where recognised under national law.

Corrective expenditure

Costs of:

public corrections;

crisis communications;

customer notifications.

Non-material damage

Potentially available in appropriate circumstances, especially for individual reputation/privacy claims.

36. Problem of Overclaiming Economic Loss

Courts may reject speculative calculations.

For example:

“Our company lost €100 million because of one AI post.”

The claimant needs evidence connecting the misinformation to the alleged loss.

A stronger case would show:

Fake news published

↓

Immediate abnormal customer cancellations

↓

Contracts explicitly terminated because of the report

↓

Revenue fell

↓

Comparable businesses did not experience the same decline

This provides stronger evidence of causation.

37. AI Fake News and Consumer Harm

AI-generated fake product information can cause:

consumers to avoid a legitimate product;

consumers to purchase dangerous products;

businesses to lose customers;

competitors to gain unfair advantages.

This may involve:

consumer-protection law;

unfair commercial practices;

unfair competition;

product liability;

defamation.

38. AI Fake News and Unfair Competition

Suppose Company A uses an AI system to generate false claims that:

“Company B's products contain dangerous chemicals.”

Customers move from B to A.

This is more than ordinary reputational harm.

Potential claims could involve:

unfair competition;

passing off or equivalent national doctrines;

misleading commercial practices;

defamation;

intentional interference with business;

damages.

39. Platform Amplification and the DSA

For very large platforms/search engines, the DSA specifically requires consideration of:

recommender systems;

advertising systems;

algorithmic amplification;

bots;

fake accounts;

coordinated manipulation;

misleading/deceptive content;

disinformation. (EUR-Lex)

This is particularly significant for AI-generated fake news because generative AI can produce enormous quantities of content at very low cost.

The risk is therefore not merely:

one false article

but:

millions of AI-generated false articles, posts, images and videos being algorithmically amplified.

40. Important Distinction: Generation vs Amplification

There are two separate causal events:

Generation

AI creates the false information.

Amplification

A platform's algorithm distributes it to a large audience.

The same defendant may be responsible for both, but they should be analysed separately.

41. AI Hallucination and Civil Liability

Suppose a user asks:

“Who founded Company X?”

The AI invents a criminal history for the founder.

The user publishes it.

Who is responsible?

Potential defendants could include:

user;

publisher;

AI provider;

platform.

But liability cannot simply be assumed.

Courts would need to determine:

applicable duty;

contractual relationship;

foreseeability;

knowledge;

control;

causation;

damage.

42. AI and Corrective Duties

A potential future doctrine may concern the duty to correct.

Suppose an AI platform is informed:

“Your system falsely stated that Company X is insolvent.”

The platform verifies the error.

If it nevertheless allows the false statement to remain prominently displayed, the failure to correct may become legally significant depending upon the applicable legal framework.

Possible remedies include:

deletion;

correction;

notice;

de-indexing;

reduced amplification;

damages.

43. Key Case-Law Table

44. Six Core Cases for Examination

If the question requires only six cases, the strongest doctrinal combination is:

1. Delfi AS v Estonia

Online intermediary responsibility.

2. Steel and Morris v United Kingdom

Defamation damages and proportionality.

3. Axel Springer AG v Germany

Balancing reputation and freedom of expression.

4. Timpul Info-Magazin and Anghel v Moldova

False allegations, defamation and compensation.

5. Kasabova v Bulgaria

Factual basis, proof and proportionality of sanctions.

6. Google Spain v AEPD

Search-engine responsibility and dissemination of personal information.

45. Future AI Fake-News Litigation

European courts are likely to encounter increasingly sophisticated disputes involving:

A. AI-generated corporate insolvency rumours

B. Deepfake CEO announcements

C. Fake central-bank announcements

D. AI-generated financial reports

E. Fake product-safety warnings

F. Synthetic celebrity endorsements

G. AI-generated medical misinformation affecting businesses

H. AI-generated false court judgments

I. Automated investment misinformation

J. AI-generated competitor attacks

K. Algorithmic amplification of fabricated stories

L. AI-generated fake customer reviews

46. Future “AI Causation” Problem

The most difficult future issue may be:

How much responsibility should be attributed to the AI system when several independent actors contributed to the damage?

Example:

Person creates false prompt

↓

AI generates false story

↓

Publisher posts it

↓

Platform recommends it

↓

Influencer reposts it

↓

Search engine ranks it highly

↓

Customers react

↓

Company loses €20 million

There may be multiple contributing causes.

Courts will therefore need to determine:

primary cause;

concurrent causes;

foreseeable consequences;

intervening acts;

contributory negligence;

allocation of liability.

47. Important Legal Distinctions

AI error ≠ automatic liability

A hallucination alone does not necessarily establish a civil wrong.

False information ≠ automatic damages

Economic loss and causation normally require proof.

Platform availability ≠ automatic platform responsibility

The legal role of the intermediary matters.

DSA violation ≠ automatic private compensation

A separate cause of action may be necessary.

AI-generated ≠ legally attributable to AI developer

Attribution depends on the relevant legal relationship and conduct.

Opinion ≠ factual allegation

But an apparently “opinion-based” AI output may contain hidden factual assertions.

48. Final Legal Formula

AI-Generated Fake News Economic Damage Claim

AI-generated false statement

↓

Publication/dissemination

↓

Identifiable claimant

↓

Wrongfulness under applicable law

↓

Attribution to defendant

↓

Causal connection

↓

Actual economic/reputational harm

↓

Quantifiable damages

↓

Potential civil liability

49. Conclusion

AI-generated fake news creates a developing European civil-liability problem because generative AI can produce false information at enormous scale and platforms can algorithmically amplify it almost instantly.

Existing European case law does not yet establish a comprehensive doctrine specifically called “AI-generated fake-news economic damage liability.” The existing authorities instead provide the building blocks: Delfi addresses intermediary responsibility; Steel and Morris addresses proportionality of defamation damages; Axel Springer, Timpul, Kasabova and Sorguç provide principles for balancing reputation against freedom of expression and distinguishing facts from opinions; while Google Spain addresses the responsibility of systems that organise and disseminate personal information. (HUDOC)

The modern DSA framework adds an important technological dimension by expressly requiring very large platforms and search engines to assess risks associated with disinformation, algorithmic amplification, recommender systems, bots and coordinated manipulation. (EUR-Lex)

Core principle

AI-generated fake news does not create liability merely because it is artificial or false. Liability depends on the applicable legal duty, the nature of the statement, attribution, wrongful dissemination, causation and legally recognised economic or reputational harm.

Exam-ready conclusion

AI Fake News + False/Harmful Information + Unlawful Dissemination + Attribution + Causation + Economic/Reputational Damage = Potential Civil Liability.