Civil Law and Artificial Intelligence Estate Administration Liability in Europe

1. Introduction

Artificial Intelligence (AI) estate administration refers to the use of AI systems in managing a deceased person's estate, including:

identifying assets and liabilities;

locating bank accounts and investments;

analysing wills and testamentary documents;

identifying beneficiaries and heirs;

valuing property and securities;

detecting debts and tax obligations;

managing digital assets;

reviewing correspondence and cloud accounts;

preparing probate documentation;

assisting executors, administrators, lawyers and notaries;

recommending distributions;

detecting possible fraud or undue influence;

administering digital assets and online accounts.

European law does not yet have a single, comprehensive “AI estate administration liability” statute. Instead, liability is constructed from several layers:

AI Act + succession law + national civil law + GDPR/data protection + contract law + professional negligence + product liability + fundamental rights.

This distinction is particularly important because the EU AI Act primarily regulates AI systems according to risk; it does not automatically make an AI provider or estate administrator liable for every loss caused by an AI output.

The AI Act expressly treats certain AI systems assisting judicial authorities or alternative dispute resolution as high-risk, because of their potential impact on fundamental rights, the rule of law and effective remedies. However, an ordinary private executor's use of AI to administer an estate is not automatically a high-risk AI system merely because it concerns inheritance. (AI Act Service Desk)

2. Meaning of AI Estate Administration Liability

The basic question is:

Who should bear the loss when an AI-assisted estate administration process produces an incorrect, unlawful or harmful result?

Possible responsible parties include:

AI developer;

AI provider;

software supplier;

estate administrator;

executor;

lawyer;

notary;

bank or financial institution;

probate professional;

beneficiary or other person who supplied incorrect information.

For example:

An executor uses AI to identify beneficiaries. The system incorrectly excludes one beneficiary, the estate is distributed, and the excluded beneficiary suffers financial loss.

Possible legal questions include:

Was the AI defective?

Did the executor negligently rely on the AI?

Was there a duty to independently verify the result?

Was the deceased's will interpreted correctly?

Was personal data lawfully processed?

Was the beneficiary's inheritance right violated?

Can the distribution be reversed?

Who must compensate the beneficiary?

3. Important Principle: AI Does Not Become the Executor

AI does not ordinarily become a separate legal person merely because it performs estate-administration tasks.

The legal responsibility normally remains with a human or legal entity such as:

executor;

administrator;

trustee;

solicitor;

notary;

bank;

professional fiduciary;

AI provider.

Therefore:

“The AI made the mistake” is normally not a complete legal defence for a professional who had a duty to supervise the system.

4. Main European Legal Framework

A. EU AI Act

The AI Act establishes a risk-based framework for AI.

For estate administration, the relevant question is not simply:

“Is this AI?”

Instead:

“What is the AI being used for, who uses it, and what risks does that particular use create?”

AI used by a private executor to organise documents may fall outside the specific high-risk categories.

By contrast, an AI system designed to assist a judicial authority in applying law or preparing judicial decisions can fall within the high-risk category concerning administration of justice. (AI Act Service Desk)

5. Succession Law Remains Primarily National

Inheritance law in Europe remains substantially governed by national law, although EU private international law is important in cross-border succession.

The EU Succession Regulation, Regulation 650/2012, is particularly relevant to international estates.

AI therefore does not replace:

national succession rules;

wills;

forced-heirship rules;

reserved shares;

testamentary capacity rules;

probate procedures;

executor powers;

rules concerning invalid wills.

An AI recommendation cannot itself determine who legally inherits property.

6. AI-Assisted Will Interpretation

One important use of AI could be interpreting:

handwritten wills;

ambiguous testamentary language;

codicils;

multiple versions of wills;

electronic documents.

Suppose a deceased person's will states:

“I leave my house to my children.”

An AI system interprets “children” as only biological children and excludes an adopted child.

That output cannot automatically determine the legal meaning of the will.

The executor or court must apply the applicable succession law.

Liability issue

If a professional simply accepts an AI interpretation without reasonable verification, possible negligence or breach-of-duty questions arise.

7. AI Identification of Heirs

AI may analyse:

family records;

genealogical databases;

civil-status documents;

wills;

correspondence.

Potential errors include:

missing an heir;

identifying the wrong heir;

confusing similarly named persons;

failing to identify an adopted child;

relying on outdated family records;

misreading a foreign-language document.

Civil consequence

A wrong identification can lead to:

incorrect distribution;

litigation;

restitution;

delay;

professional liability;

damages.

8. AI Asset Discovery

Modern estates may contain:

bank accounts;

cryptocurrency;

shares;

NFTs;

intellectual-property rights;

online businesses;

domain names;

cloud storage;

insurance policies;

digital wallets.

AI can be used to search documents and identify assets.

But an AI system may:

overlook an account;

incorrectly classify an asset;

double-count assets;

misvalue property;

fail to detect cryptocurrency;

identify someone else's account as belonging to the estate.

The administrator's duty of care therefore becomes central.

9. AI Valuation of Estate Assets

AI may estimate:

real-estate values;

securities;

businesses;

collectibles;

intellectual property;

cryptocurrency.

A wrong valuation can affect:

inheritance distribution;

tax;

forced-heirship calculations;

equalisation among beneficiaries;

sale decisions.

Example

Estate property worth €1 million is incorrectly valued by an AI system at €600,000.

The executor distributes the estate on that basis.

Later the actual value is established.

Potential claims may concern:

negligent administration;

breach of fiduciary duty;

professional negligence;

restitution;

damages.

10. AI and Digital Inheritance

Digital inheritance creates one of the most interesting connections between succession law and AI.

A deceased person may leave:

email accounts;

social-media accounts;

cloud files;

photographs;

cryptocurrency wallets;

digital businesses;

AI-generated content;

subscription accounts.

A major European authority is the German Facebook inheritance case.

11. Case 1 — Bundesgerichtshof, Facebook Account Inheritance

BGH, III ZR 183/17, 12 July 2018

This is one of the most important European cases for digital estate administration.

The German Federal Court of Justice considered whether the parents of a deceased minor could obtain access to her Facebook account.

The Court held, in substance, that the contractual position concerning the Facebook account could pass to the heirs in the same way as other contractual positions, and that the heirs were entitled to access the account.

The case is particularly significant because the Court rejected the idea that the digital account should simply disappear from the estate.

The decision also dealt with telecommunications secrecy and data protection issues. German case-law materials explain that the GDPR concerns living natural persons, not the deceased as such. (EUR-Lex)

AI-estate significance

An AI estate administrator may be instructed to:

identify digital accounts;

organise digital communications;

classify digital assets;

retrieve information relevant to inheritance.

The case demonstrates that digital assets and account relationships can form part of an estate.

It is therefore a foundational case for AI-assisted digital estate administration.

12. Case 2 — Google Spain

Google Spain SL and Google Inc. v AEPD and Mario Costeja González

C-131/12, CJEU

The CJEU addressed responsibility for processing personal information by search engines.

The case established important principles concerning:

personal data;

search-engine processing;

individual rights;

removal of certain search results;

balancing privacy and information interests.

AI-estate relevance

AI estate-management systems may use:

search engines;

automated information retrieval;

public databases;

online profiles.

An AI administrator searching for heirs or assets may therefore encounter information concerning living persons.

The estate administrator cannot assume that all publicly discoverable information can automatically be collected and processed for any purpose.

13. Case 3 — SCHUFA Holding (Scoring)

C-634/21, CJEU, 7 December 2023

The CJEU held that the automated establishment of a probability value can constitute automated individual decision-making where a third party strongly relies on that score in establishing, implementing or terminating a contractual relationship. (EUR-Lex)

Estate-administration relevance

Imagine an AI system assesses beneficiaries or creditors and produces a risk score that an administrator automatically uses to decide:

whether a beneficiary is genuine;

whether a creditor should be paid;

whether a claimant should receive an inheritance;

whether a person should be excluded from further investigation.

The SCHUFA reasoning demonstrates why an apparently “advisory” algorithm can become legally important if the human decision-maker effectively relies upon it.

Principle

A label such as “recommendation” does not necessarily prevent automated decision-making rules from becoming relevant.

14. Case 4 — CK v Magistrat der Stadt Wien

C-203/22, CJEU, 27 February 2025

This is an important recent development concerning algorithmic decision-making and transparency.

The CJEU held that “meaningful information about the logic involved” requires relevant information concerning the procedure and principles actually used in automated decision-making. A merely complex mathematical formula is not sufficient. (EUR-Lex)

The Court emphasised that the explanation must be:

concise;

transparent;

intelligible;

accessible.

(EUR-Lex)

Estate significance

Suppose an AI estate system recommends that:

Beneficiary A receives 60%;

Beneficiary B receives 20%;

Beneficiary C receives nothing.

If the system's output significantly affects a living person's rights, questions may arise concerning:

what data were used;

how the system reached the result;

whether incorrect information was used;

whether the result can be challenged.

This is particularly important where AI is used to make or substantially influence legal or financial decisions.

15. Case 5 — Österreichische Post

C-300/21, CJEU, 4 May 2023

The CJEU examined compensation under GDPR Article 82.

The Court distinguished between:

infringement of the GDPR; and

compensable damage.

The existence of an infringement does not by itself mean that damages automatically follow. The claimant still needs the legally required damage and causal relationship.

Estate relevance

An AI estate administrator could unlawfully process information concerning:

beneficiaries;

family members;

creditors;

executors;

business partners.

But a GDPR infringement and a civil compensation claim remain distinct questions.

This is an important principle for AI estate litigation.

16. Case 6 — Natsionalna agentsia za prihodite

C-340/21, CJEU, 14 December 2023

This case concerned personal-data security following a cyberattack.

The CJEU considered:

security obligations;

technical and organisational measures;

controller responsibility;

potential non-material damage.

AI-estate relevance

Estate administrators increasingly hold extremely sensitive information:

wills;

bank details;

tax records;

passwords;

family relationships;

cryptocurrency information.

If an AI estate platform is hacked and beneficiary information is exposed, potential liability may involve:

GDPR;

contractual duties;

professional negligence;

cybersecurity obligations.

The case therefore provides an important analogy for AI-based estate-management platforms.

17. Case 7 — Österreichische Datenschutzbehörde and CRIF

C-487/21, CJEU, 4 May 2023

The CJEU emphasised the importance of the right of access to personal data and meaningful information concerning processing.

The Court linked access rights to the ability of data subjects to exercise rights such as:

rectification;

erasure;

restriction;

objection;

legal action;

compensation.

This principle is particularly important for AI estate systems because an individual may need to discover what personal information an automated system used.

18. Case 8 — Boston Scientific Medizintechnik

Joined Cases C-503/13 and C-504/13, CJEU

This case concerned defective medical devices rather than estate administration.

It is nevertheless important as an analogical product-liability authority.

The CJEU recognised that a product may be defective where there is a potential safety problem affecting products in the same series, without requiring proof that every individual product has already malfunctioned.

AI-estate relevance

Suppose an estate-management software version contains a systemic error that:

incorrectly calculates beneficiaries' shares;

systematically excludes a class of heirs;

systematically miscalculates estate values.

The case illustrates why product liability can sometimes focus on systemic safety defects, rather than requiring proof that every individual use was defective.

It is not a direct inheritance case.

19. Direct Estate Cases vs AI Cases

This distinction is essential.

There is currently far less European reported case law directly concerning AI-controlled estate administration than there is concerning automated decision-making, digital inheritance and data protection. Therefore, it would be legally inaccurate to present all of these authorities as direct “AI inheritance” cases.

20. AI and the Executor's Duty of Care

The central civil-law question is often:

Did the executor use reasonable care when relying on AI?

The answer depends on:

professional status;

complexity of the estate;

foreseeable risks;

instructions supplied by the AI provider;

reliability of the system;

importance of the decision;

availability of human verification;

national succession law.

For example, relying on AI to alphabetically organise thousands of documents may be relatively low-risk.

Relying on AI alone to decide whether someone is legally an heir is much more consequential.

21. Human Verification

A prudent AI-assisted estate process should ordinarily include human verification of:

identity;

testamentary documents;

heirship;

asset ownership;

debts;

tax liabilities;

valuation;

distribution calculations.

This is particularly important because an AI system may produce a plausible but legally incorrect answer.

22. AI and Testamentary Documents

AI may assist with:

OCR of handwritten wills;

translation;

document comparison;

detection of inconsistent clauses;

chronology;

document classification.

But AI should not automatically determine:

testamentary capacity;

undue influence;

authenticity;

validity;

applicable succession law.

Those are legal questions.

23. AI and Undue Influence

A particularly sensitive issue arises if AI analyses a deceased person's communications to determine whether a will was influenced by another person.

Potential evidence could include:

emails;

messages;

financial transactions;

medical records where lawfully available;

correspondence;

drafts of wills.

But automated inference may be unreliable.

A system might wrongly interpret:

“I want my daughter to handle everything”

as evidence of coercion when the statement was actually voluntary.

The resulting estate dispute could involve:

inheritance litigation;

privacy;

confidentiality;

professional negligence;

evidence law.

24. AI and Testamentary Capacity

AI could potentially be used to analyse historical documents or communications relevant to capacity.

However, AI-generated conclusions about mental capacity should not automatically be treated as legally determinative.

The legal question normally requires application of the relevant national law and evidence.

A professional who treats an AI prediction as a substitute for legally required evidence could potentially face liability.

25. AI and Beneficiary Identification

This is one of the highest-risk practical applications.

An AI system could compare:

birth records;

marriage records;

adoption records;

wills;

family databases.

Potential mistakes include:

excluding an adopted child;

including a deceased person;

confusing half-siblings;

missing a later will;

failing to identify a foreign heir.

Legal consequence

The estate could be distributed incorrectly.

The rightful beneficiary could seek:

recovery of inherited property;

restitution;

damages;

correction of probate proceedings;

professional liability.

26. AI and Digital Assets

Digital assets present unique challenges.

Cryptocurrency

The administrator may need to identify:

wallets;

private keys;

exchange accounts.

Digital businesses

The estate may contain:

online stores;

domain names;

advertising accounts;

platform contracts.

AI-generated intellectual property

The estate may contain:

AI-generated works;

prompts;

datasets;

model-related contracts.

Cloud accounts

The administrator may need to establish contractual and succession rights.

The German Facebook inheritance case demonstrates the importance of treating digital account relationships within the broader succession framework. (EUR-Lex)

27. Data Protection and Deceased Persons

An important distinction must be made.

The GDPR protects natural persons concerning personal data. It does not generally regulate personal data of deceased persons in the same manner as living persons.

The German Facebook inheritance case is particularly significant because the German Federal Court of Justice treated the deceased person's account as part of the inheritance while addressing the separate data-protection position of living persons. (EUR-Lex)

Therefore:

An AI administrator cannot assume that because the person is deceased, all information connected with the estate becomes unrestricted.

Information may concern living:

heirs;

family members;

correspondents;

employees;

business partners.

Their rights remain relevant.

28. AI and Confidentiality

Estate administrators often possess confidential information.

AI systems may transmit information to:

cloud providers;

model providers;

subcontractors;

analytics services.

This creates questions concerning:

confidentiality;

professional secrecy;

data processing;

cybersecurity;

contractual restrictions.

A solicitor who uploads an entire confidential estate file into an external AI system may therefore create separate professional and data-protection issues.

29. AI and Cross-Border Estates

European estates frequently involve several jurisdictions.

Example:

deceased domiciled in France;

property in Spain;

bank account in Germany;

cryptocurrency exchange elsewhere;

heirs in Italy and Belgium.

AI may be used to coordinate the estate.

But the administrator still needs to determine:

applicable succession law;

jurisdiction;

property law;

tax law;

contractual law;

data-protection requirements.

AI cannot itself choose the governing law merely because its model predicts a particular legal outcome.

30. AI and Estate Tax

AI may calculate:

inheritance tax;

estate tax;

capital gains;

property valuation;

deductions.

A computational error can cause:

underpayment;

overpayment;

penalties;

interest;

loss to beneficiaries.

Where a professional relies on AI, the question may become whether the professional had a duty to verify the tax calculation.

31. AI and Estate Distribution

Consider:

Estate = €3 million.

Three beneficiaries are entitled to different shares.

AI incorrectly calculates:

Beneficiary A: €1.5m;

Beneficiary B: €1m;

Beneficiary C: €500,000.

The correct distribution should be different.

Potential liability could arise from:

incorrect software;

incorrect data entry;

failure of the administrator to verify;

incorrect interpretation of the will;

incorrect legal advice.

The court must therefore determine where the causal error occurred.

32. AI Provider Liability

The AI provider may potentially be responsible where:

the software is defective;

the system fails contractual specifications;

safety information is inadequate;

the provider misrepresents functionality;

the system has a systematic calculation defect.

The EU's new product-liability framework is increasingly important because software is expressly brought within the concept of a product.

However, not every bad AI output automatically proves a legally compensable product defect.

33. Administrator Liability

The estate administrator may be liable where the administrator:

blindly relies on AI;

fails to verify obvious errors;

ignores contradictory evidence;

fails to preserve estate assets;

distributes assets prematurely;

uses unlawful data;

discloses confidential information.

This is generally a question of the administrator's applicable professional, fiduciary, contractual or tort duties under national law.

34. Causation

AI estate cases can involve several possible causes.

Example:

AI incorrectly identifies heir → executor accepts result → executor distributes estate → beneficiary loses inheritance.

Possible causes include:

AI defect;

bad source data;

executor's failure to verify;

ambiguous will;

incorrect legal interpretation.

The claimant must identify the legally relevant causal chain.

35. Evidence in AI Estate Litigation

Important evidence may include:

AI prompts;

AI outputs;

system logs;

model version;

training or reference material;

estate database;

document history;

human approvals;

audit records;

email correspondence;

valuation records;

beneficiary calculations.

The CK case is particularly useful in demonstrating the broader European importance of meaningful explanations for automated decision-making. The CJEU held that the explanation must concern the procedure and principles actually applied and must be understandable rather than merely providing a complicated algorithmic formula. (EUR-Lex)

36. AI Hallucinations in Estate Administration

A serious practical problem is hallucination.

An AI system may invent:

a beneficiary;

a legal rule;

an asset;

a debt;

a court case;

a provision of a will.

For estate administration, this can be extremely dangerous because an apparently confident answer may be mistaken.

Legal significance

A professional cannot necessarily argue:

“The AI told me it was correct.”

The relevant question may instead be whether reasonable professional verification was required.

37. AI and Fraud Detection

AI can identify:

suspicious transfers;

unusual beneficiary changes;

forged documents;

unexplained transactions.

But false positives can also cause harm.

For example, AI identifies a legitimate beneficiary as a fraud risk.

If the executor refuses payment solely because of the AI score, the beneficiary could challenge the decision.

The SCHUFA reasoning is relevant by analogy because the legal significance of automated scoring can depend on the extent to which a human decision-maker actually relies upon it. (EUR-Lex)

38. AI and Human Rights

Estate administration can involve:

privacy;

family life;

property rights;

equality;

access to justice.

AI systems must therefore not be treated as purely financial tools.

The European data-protection jurisprudence stresses the relationship between data protection, private life and fundamental rights. (EUR-Lex)

39. Liability Matrix

40. Relationship Between the AI Act and Estate Liability

The correct analytical model is:

AI Act

Was the AI system lawfully developed/deployed?

↓

Succession law

Who legally inherits?

↓

Contract/professional law

What duties did the administrator, lawyer, notary or service provider owe?

↓

GDPR

Was personal data lawfully processed?

↓

Product liability

Was the software/product defective?

↓

Tort/delict

Did unlawful or negligent conduct cause damage?

↓

Remedy

Can the distribution be reversed and/or damages awarded?

41. Six Core Legal Principles

1. AI is an administrative tool, not the legal executor

The human/legal entity responsible for the estate generally remains responsible for fulfilling its legal duties.

2. AI Act compliance does not automatically eliminate civil liability

Compliance with AI regulation does not necessarily mean that a separate contractual or tort duty was satisfied.

3. AI Act violation does not automatically establish damages

The claimant still needs an appropriate civil-liability basis and the legally required damage and causation.

4. Digital assets can form part of an estate

The German Facebook case is especially important here. (EUR-Lex)

5. Automated decision-making can trigger transparency rights

SCHUFA and CK demonstrate the importance of meaningful information concerning automated decisions. (EUR-Lex)

6. Data protection remains relevant to living persons connected with the estate

The death of the testator does not remove GDPR concerns concerning living heirs and other individuals.

42. Important Case-Law Revision Table

43. Exam Answer Structure

For an examination problem involving AI estate administration, write in this order:

Step 1 — Identify the AI use

Was AI used for:

heir identification;

will interpretation;

asset discovery;

valuation;

distribution;

tax;

digital-account administration?

Step 2 — Identify the legal actor

Determine whether the defendant is:

executor;

administrator;

lawyer;

notary;

bank;

AI provider;

software manufacturer.

Step 3 — Apply succession law

Determine who legally inherits.

Step 4 — Apply AI Act

Determine whether the particular AI use falls within a regulated risk category.

Step 5 — Apply GDPR

Consider processing of information concerning living heirs and other persons.

Step 6 — Consider product liability

Ask whether defective software caused damage.

Step 7 — Consider contract/professional negligence

Examine the administrator's duties.

Step 8 — Establish causation

Connect the AI error to the actual loss.

Step 9 — Determine remedy

Possible remedies may include:

correction of distribution;

restitution;

injunction;

damages;

access to information;

rectification;

deletion;

contractual remedies.

44. Conclusion

Artificial Intelligence estate administration in Europe is governed by a layered civil-law framework rather than a single AI inheritance-liability regime.

The AI Act supplies regulatory standards; succession law determines inheritance; GDPR regulates personal-data processing; national contract, tort and professional-liability rules determine many compensation questions; and product-liability rules increasingly apply to software and AI products.

The most important direct European authority is the German Federal Court of Justice's Facebook inheritance decision, which establishes the significance of digital accounts in succession. The CJEU's more recent SCHUFA and CK decisions provide important principles for automated decision-making, transparency and meaningful explanations, while Österreichische Post, CRIF, and Natsionalna agentsia provide the data-protection liability framework. (EUR-Lex)

The central principle can therefore be stated simply:

AI may assist in administering an estate, but it does not replace the legal duties of the executor, administrator, lawyer, notary or other responsible human/legal entity.

Where AI incorrectly identifies an heir, misinterprets a will, misvalues property, loses digital assets or unlawfully processes information, liability must be determined by examining the AI system, the human decision-maker, the applicable succession rules, the source of the error, the resulting damage and the causal connection between them.