Civil Law and Autonomous Factory Shutdown Loss Compensation in Europe

1. Meaning

Autonomous factory shutdown loss compensation concerns civil claims arising when an automated or AI-controlled industrial system independently, or with limited human intervention, shuts down a factory or production line and causes economic loss.

Examples include:

AI-controlled production lines stopping unexpectedly.

Autonomous safety systems triggering an unnecessary shutdown.

Predictive-maintenance software incorrectly identifying a critical failure.

Industrial robots or PLC/SCADA systems issuing an erroneous shutdown command.

Cybersecurity systems automatically isolating an entire factory.

AI energy-management systems disconnecting production equipment.

Autonomous quality-control systems stopping production because of a false defect detection.

Software updates causing machinery to become unavailable.

An autonomous system continuing to operate despite detecting a dangerous condition and subsequently causing a shutdown.

There is no settled European civil-liability doctrine specifically named “autonomous factory shutdown liability.” The claim normally has to be constructed from contract law, product liability, tort/delict, professional negligence, machinery/software liability, causation and damages principles.

2. Main Legal Issue

The central question is:

Who should compensate the factory operator for losses caused by an autonomous shutdown, and what losses are legally recoverable?

Potential defendants include:

manufacturer of the machinery;

software/AI developer;

system integrator;

maintenance contractor;

cybersecurity provider;

cloud/remote-control provider;

factory operator itself;

employee or supervisor responsible for configuration;

supplier of a defective component;

several parties jointly.

The fact that the shutdown was performed autonomously does not ordinarily make the AI system itself a separate legal person or automatically make it liable. Liability generally remains attributable to a human or corporate legal entity.

3. Principal Causes of Action

A. Contractual liability

A factory will often have contracts with:

equipment manufacturers;

automation suppliers;

software developers;

maintenance companies;

system integrators;

cloud providers.

The contract may contain:

uptime guarantees;

availability guarantees;

service-level agreements;

maintenance obligations;

response-time obligations;

software-update obligations;

cybersecurity obligations;

production guarantees;

performance warranties;

business-continuity obligations.

If an autonomous shutdown results from breach of such obligations, contractual damages may be claimed.

Example

A manufacturer guarantees 99.9% availability of an automated production system. Its software incorrectly shuts down the production line for five days.

The factory may claim:

repair expenses;

emergency replacement costs;

additional labour;

lost production;

potentially lost profits, subject to applicable national law and contractual limitations.

4. Product Liability

A major route is defective-product liability.

Historically, the EU Product Liability Directive 85/374/EEC created a harmonised strict-liability regime for defective products.

The modern Product Liability Directive (EU) 2024/2853 substantially expands the framework to modern digital products, including software-related situations.

This is particularly significant for autonomous factories because modern industrial equipment can consist of:

physical machine + embedded software + AI model + cloud connection + cybersecurity system + continuous updates.

A defect can therefore be:

hardware defect;

software defect;

design defect;

inadequate instructions;

inadequate warning;

unsafe update;

cybersecurity vulnerability;

defective AI decision logic.

However, a major qualification is important: pure economic losses such as lost production or lost profits do not automatically fall within every product-liability regime. The precise recoverability depends upon the applicable version of EU product-liability law and national law, as well as whether property damage or personal injury is involved.

5. Six Important European Case Laws

Case 1 — Boston Scientific Medizintechnik GmbH v AOK Sachsen-Anhalt

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

Principle

The CJEU considered when a product may be considered defective where products belonging to the same production series present an abnormal risk.

The Court adopted a strong safety-oriented approach: a product can be defective where it does not provide the safety which a person is entitled to expect, taking account of all circumstances.

Application to autonomous factories

Suppose an industrial controller manufactured with the same software version contains a dangerous shutdown defect.

If one unit reveals a systemic safety problem, evidence concerning the wider production series can become important.

Relevant evidence could include:

previous shutdowns;

software logs;

error codes;

engineering alerts;

manufacturer recalls;

internal safety testing;

previous customer complaints.

Importance

The case supports the proposition that systemic product risks can be relevant to defect analysis, rather than examining only one individual machine in isolation.

6. Case 2 — Sanofi Pasteur SA v France

CJEU, Case C-621/15

Principle

The CJEU examined causation in defective-product litigation, including situations where scientific certainty concerning causation may be difficult to establish.

The Court recognised the importance of national evidentiary rules while maintaining the requirements of the EU product-liability framework.

Application

Autonomous factory shutdown cases can present similar causation difficulties.

For example:

Did the AI software cause the shutdown, or did an underlying electrical fault cause both the AI alarm and the shutdown?

Possible alternative causes include:

power fluctuations;

network failure;

human configuration;

defective sensors;

malware;

hardware deterioration;

incorrect software update;

external grid failure.

Expert technical evidence may therefore be crucial.

Importance

Causation cannot simply be assumed from the fact that an autonomous system issued the shutdown command.

7. Case 3 — Veedfald v Århus Amtskommune

CJEU, Case C-203/99

Principle

The CJEU interpreted the scope of defective-product liability and emphasised the significance of the product and its intended use.

The case concerned a product used in the course of medical treatment and the application of the EU product-liability regime.

Application to autonomous industrial systems

The same conceptual distinction is useful in factory disputes:

Was the machine itself defective?

Was the software defective?

Was the system improperly installed?

Was the system used contrary to its intended purpose?

Did the manufacturer provide adequate instructions?

An autonomous factory system may therefore require analysis of the whole operational environment, rather than only the AI algorithm.

8. Case 4 — O'Byrne v Sanofi Pasteur MSD Ltd

CJEU, Case C-127/04

Principle

The CJEU examined when a product is regarded as having been “put into circulation” for purposes of product liability.

The Court treated the concept by examining when the product leaves the producer's manufacturing process and enters the marketing process.

(Eur-Lex)

Application

This becomes relevant where a factory system has several participants:

Manufacturer → subsidiary → system integrator → factory operator

Suppose a software-controlled machine fails after being modified by an integrator.

Questions may include:

Who placed the system into circulation?

Who modified it?

Who supplied the defective software?

Was the defect present before integration?

Was the later modification responsible?

Importance

Autonomous factory claims can therefore require careful identification of the responsible participant in a complex technological supply chain.

9. Case 5 — Dutrueux

CJEU, Case C-495/10

Principle

The case concerned damage resulting from defective equipment used in healthcare and the relationship between EU product-liability rules and national liability regimes.

The CJEU recognised that the EU product-liability regime does not necessarily prevent Member States from maintaining other liability rules within their own legal systems.

(Eur-Lex)

Application

This is particularly important for factory shutdown claims.

A claimant may potentially have several legal routes:

Product liability

plus

contractual liability

plus

national tort/delict liability

depending upon the facts and the applicable national law.

Thus, failure of a product-liability claim does not necessarily mean that all civil remedies disappear.

10. Case 6 — Greek State v Inter-Kom

CJEU, Case 71/87

Principle

The CJEU discussed force majeure, describing circumstances involving abnormal and unforeseeable events beyond the person's control whose consequences could not have been avoided despite appropriate care.

The Court specifically recognised that an interruption in electricity supply could potentially constitute force majeure depending upon the circumstances. (Eur-Lex)

Application to autonomous factories

This is highly relevant where an autonomous shutdown is triggered by an external event.

For example:

unexpected grid failure;

extreme weather;

external telecommunications failure;

major cyberattack;

government emergency order;

extraordinary infrastructure failure.

But an ordinary software defect or equipment failure will not automatically qualify as force majeure merely because the system operated autonomously.

The question is whether the event was genuinely outside the responsible party's control and unavoidable despite reasonable care.

11. Case 7 — Case 314/06, Commission v Spain

The CJEU's force-majeure jurisprudence has also emphasised that the event must be abnormal, unforeseeable and objectively outside the relevant person's sphere of responsibility, despite the exercise of due care. (Eur-Lex)

Application

A manufacturer cannot necessarily argue:

“The AI unexpectedly shut down the factory, therefore the event was force majeure.”

That argument is weak where the shutdown resulted from:

foreseeable software failure;

inadequate testing;

inadequate cybersecurity;

known sensor problems;

failure to install safety patches;

inadequate maintenance.

Autonomous behaviour is not synonymous with unforeseeability.

12. Case 8 — Systran SA v Commission

General Court, Case T-19/07

Principle

The Court stressed the requirement of a sufficiently direct causal connection between the wrongful conduct and the damage claimed.

It rejected parts of a claim where the alleged loss could not be sufficiently directly attributed to the relevant conduct. (Eur-Lex)

Application

This principle is especially important for factory shutdown claims.

Suppose an AI system shuts down a factory for three days.

The claimant might calculate:

€500,000 lost production;

€200,000 lost profit;

€100,000 customer penalties;

€50,000 reputational loss;

€300,000 future business losses.

The claimant must establish a sufficiently convincing causal chain.

13. What Losses Can Be Claimed?

A factory operator could potentially claim several categories.

14. Lost Profits and Business Interruption

This is often the most difficult part of an autonomous shutdown claim.

A factory cannot simply say:

“The factory was closed for ten days, therefore ten days of normal revenue should be awarded.”

The court may examine:

historical production;

normal profit margins;

cancelled orders;

replacement production;

inventory;

market conditions;

seasonal fluctuations;

alternative suppliers;

mitigation;

whether production would actually have occurred.

European damages jurisprudence recognises that economic loss and lost profit can be compensable in appropriate circumstances, but the claimant must establish the loss and causal relationship rather than rely on speculation. (Eur-Lex)

15. Autonomous Shutdown and Causation

A useful legal model is:

System defect

↓

Incorrect autonomous decision

↓

Factory shutdown

↓

Production interruption

↓

Commercial consequences

↓

Financial loss

The claimant must establish each important link.

For example:

defective temperature sensor → AI falsely detects dangerous overheating → automatic shutdown → production stops → contractual deliveries missed → identifiable financial loss.

This is considerably stronger than:

AI malfunction → company suffered general financial difficulties.

16. Manufacturer Liability

A manufacturer may face liability where:

the machine was defectively designed;

the software contained a dangerous error;

safety logic was inadequate;

the manufacturer failed to provide warnings;

a known vulnerability was not corrected;

a defective update caused the shutdown;

cybersecurity safeguards were inadequate;

the system's autonomous behaviour exceeded reasonable safety expectations.

The manufacturer may defend itself by arguing:

no defect existed;

defect was introduced after delivery;

improper use;

unauthorised modification;

third-party cyberattack;

force majeure;

intervening cause;

contractual limitation;

absence of causation.

17. AI/Software Developer Liability

The software developer becomes particularly important where the physical machinery was functioning correctly but the AI made the shutdown decision.

Possible allegations include:

defective algorithm;

inadequate training data;

incorrect threshold;

false-positive detection;

inadequate testing;

failure to account for foreseeable operating conditions;

unsafe update;

inadequate documentation;

inadequate monitoring;

cybersecurity weakness.

The legal classification depends heavily on the contractual relationship and applicable national law.

18. System Integrator Liability

The integrator may be responsible where it combines:

machine + sensors + AI + PLC + cloud + factory network.

For example:

Each component individually works correctly, but the integrator configures the AI threshold incorrectly.

The resulting shutdown may therefore be attributable to the integration process, rather than the original manufacturer.

19. Factory Operator's Own Responsibility

The factory owner may also bear some or all responsibility if it:

ignored warnings;

failed to maintain equipment;

disabled safety features;

installed unauthorised software;

failed to update systems;

used machinery outside specifications;

improperly configured autonomous controls;

ignored repeated warning signals.

This can lead to:

contributory negligence;

reduction of damages;

contractual exclusion;

complete defence in some circumstances.

20. Force Majeure

A major issue is whether an autonomous shutdown was caused by an extraordinary external event.

Possible force majeure

unprecedented natural disaster;

extraordinary grid failure;

exceptional government intervention;

major external cyberattack;

extraordinary telecommunications failure.

Usually weaker force-majeure arguments

ordinary software bugs;

predictable hardware failures;

inadequate maintenance;

known cybersecurity vulnerabilities;

defective AI training;

poor system integration.

The CJEU's established approach requires circumstances that are abnormal and unforeseeable, outside the relevant party's control, and unavoidable despite due care. (Eur-Lex)

21. Contractual Limitation Clauses

Industrial automation agreements frequently contain:

liability caps;

exclusion of consequential losses;

exclusion of lost profits;

service-level remedies;

force-majeure clauses;

warranty periods;

exclusive-remedy clauses.

Their validity and interpretation depend substantially on national law.

Courts may examine:

whether the clause was incorporated;

whether it was clearly drafted;

bargaining power;

mandatory statutory liability;

consumer/SME protection where applicable;

gross negligence or intentional misconduct;

whether the clause excludes legally non-excludable liability.

22. Evidence in Autonomous Shutdown Litigation

Evidence is particularly important because the system itself may have generated the relevant information.

Important evidence includes:

Technical evidence

system logs;

PLC logs;

sensor data;

AI model outputs;

error codes;

timestamps;

system architecture;

configuration files;

software versions;

update history.

Corporate evidence

maintenance records;

incident reports;

engineering emails;

previous shutdown reports;

risk assessments;

cybersecurity reports.

Commercial evidence

production records;

sales orders;

cancelled contracts;

customer penalties;

inventory records;

profit calculations.

23. The Black-Box Problem

An autonomous system may produce an instruction:

“Emergency shutdown.”

But the factory operator may not know:

Why?

This creates a major litigation problem.

The claimant may need to establish:

what input the system received;

what model/version was operating;

what threshold was used;

whether the system was modified;

whether the decision was predictable;

whether a human could override it;

whether previous warnings existed.

This is where modern AI-related transparency and evidence rules may become increasingly important, although they do not create a single general European damages regime for autonomous factories.

24. Product Liability vs Contract Liability

25. Special Problem: Cybersecurity Shutdown

An autonomous factory may shut itself down because its cybersecurity system detects suspicious network activity.

Suppose:

External attacker → abnormal network traffic → AI security system → automatic isolation → factory shutdown.

Three separate questions arise:

1. Was the cybersecurity system defective?

If yes, manufacturer/provider liability may arise.

2. Was the cyberattack force majeure?

Possibly, depending upon its character and foreseeability.

3. Was the factory operator adequately protected?

If the operator failed to implement reasonable cybersecurity measures, contributory responsibility may arise.

Thus, cyberattack does not automatically eliminate civil liability.

26. Autonomous Shutdown Caused by Software Update

A particularly important modern scenario is:

Manufacturer releases update → update changes AI decision logic → false alarm → autonomous shutdown.

Potential defendants include:

original manufacturer;

software developer;

update provider;

system integrator;

maintenance contractor.

Evidence of previous incidents following the update could be highly relevant to:

defect;

foreseeability;

knowledge;

causation;

negligence;

mitigation.

The modern EU Product Liability Directive is particularly significant here because the EU's updated product-liability framework expressly accommodates modern software and products affected by subsequent modifications and updates.

27. Direct and Analogical Case Law

The cases concerning product liability and force majeure are not themselves cases about AI-controlled factories. They provide the established European principles that can be applied by analogy to autonomous industrial systems.

28. Hypothetical Example

Assume an automobile factory uses an autonomous production-control system.

The AI detects what it believes to be a dangerous robotic-arm condition and shuts down the entire plant for seven days.

Investigation shows:

sensor was functioning correctly;

AI software incorrectly interpreted the sensor;

manufacturer had previously received similar complaints;

software update had not been installed;

factory had followed its normal maintenance schedule.

The factory claims:

€400,000 repair costs;

€800,000 lost production;

€300,000 lost profits;

€100,000 customer penalties.

Legal analysis

Step 1 — Defect

Was the autonomous system insufficiently safe?

Step 2 — Attribution

Manufacturer, developer or integrator?

Step 3 — Causation

Did the AI error actually cause the shutdown?

Step 4 — Damage

Which financial losses are sufficiently proven?

Step 5 — Mitigation

Could production have been transferred elsewhere?

Step 6 — Contract

Was there an uptime/SLA guarantee?

Step 7 — Defences

Was the system improperly used or maintained?

Step 8 — Force majeure

Was the event genuinely outside the defendant's control?

29. Important Legal Principle

The most important distinction is:

Autonomous decision-making does not itself create autonomous legal liability.

The legal system generally looks behind the automated decision to determine:

Who designed it?
Who supplied it?
Who configured it?
Who operated it?
Who maintained it?
Who controlled the update?
Who breached the relevant duty?

That entity can then potentially bear civil responsibility.

30. Conclusion

Autonomous factory shutdown compensation in Europe is best understood as a multi-layer civil-liability problem, rather than as a separate established cause of action.

The principal legal routes are:

contractual breach;

product liability;

tort/delict;

professional negligence;

software/AI defect;

cybersecurity liability;

causation and loss-of-profit principles;

force majeure;

contributory negligence;

statutory and contractual limitations.

The strongest claims will normally require a technically demonstrable chain:

defective autonomous system → wrongful/unsafe shutdown → interruption of production → proven financial loss → sufficiently direct causal connection.

The European authorities discussed above establish the building blocks for that analysis, although there is not yet a mature body of CJEU case law specifically dealing with an AI-controlled factory autonomously shutting down production and awarding compensation for the resulting business-interruption losses. (Eur-Lex)

Exam / Revision Keywords

Autonomous shutdown — Industrial AI — Factory automation — Product defect — Software defect — Machine liability — System integrator — Contract breach — SLA — Product Liability Directive — Causation — Lost production — Lost profits — Business interruption — Force majeure — Cyberattack — Software update — AI error — Sensor failure — Mitigation — Contributory negligence — Liability cap — Evidence logs — Black-box system — Defective product — Economic loss — Direct causal link.