PRINCIPLE OF RESPONSIBILITY OF NEW INTELLIGENCES II - english translation - Foundations, Analogies, and Transcending Classical Regimes

 “No intelligence can exist without responsibility.

The only question is where it must be legally attached.”

“No intelligence can exist without responsibility.
The only question is where it must be legally attached.”

 

INTRODUCTION

With the emergence of so-called new forms of intelligence, it becomes possible to distinguish three forms of intelligence.

The first pertains to a supreme intelligence, an organizing principle of reality as a whole, whose manifestations structure both the physical world and abstract systems.

The second corresponds to human intelligence, a faculty of perception, interpretation, and organization, enabling humans to interact with this order without necessarily grasping it in its entirety.

The third, more recent in its operational manifestations, is what is referred to as new intelligence. It does not result from a creation ex nihilo by humans, but from a technical organization of a pre-existing principle, materialized within computational systems.

Whatever its level, intelligence remains immaterial in essence and is revealed only through its effects.

However, legal regimes of liability were designed to govern behaviors attributable to identifiable subjects capable of answering for damage.

A tension therefore arises: how can regimes based on human imputability be applied to systems whose functioning relies on an immaterial and partially autonomous intelligence?

Part of legal doctrine proposes the creation of an autonomous liability regime, or even the recognition of a legal personality specific to artificial intelligence.

However, such an approach would introduce a rupture in the foundations of liability law by dissociating the effect from its author.

This article proposes a different approach.

It is not a matter of creating new principles, but of reconstructing the application of existing principles in light of a fundamental distinction between forms of intelligence.

This reconstruction leads to the identification of a guiding principle:

any organized immaterial intelligence implies a material responsibility attributable to its organizer.

 

I – SUPREME INTELLIGENCE: THE STRUCTURAL IMPOSSIBILITY OF LIABILITY

Supreme intelligence is defined as that above which no other form of intelligence can be conceived.

In this perspective, any attempt to attribute liability encounters a conceptual impossibility.

Liability presupposes the existence of a superior or at least equivalent authority capable of judging, qualifying a fault, and organizing reparation.

In the absence of such an authority, liability becomes inoperative.

The principle according to which “he who can do more can do less” finds a particular application here: an absolute intelligence cannot be subject to a regime designed for limited intelligences.

It follows that:

• no action for reparation can be envisaged,
• no fault can be legally qualified,
• no damage can be attributed in the sense of positive law.

This impossibility is not contingent, but structural.

It highlights that liability can only exist between entities situated in a relationship of equivalence or normative hierarchy.

Regulatory mechanisms can therefore only operate between lower-level intelligences, that is, within the domain of human law.

 

II – HUMAN INTELLIGENCE: THE CLASSICAL FOUNDATION OF LIABILITY

Human intelligence operates within a social framework structured by legal norms.

The cardinal principle of liability is well established:

any act causing damage obliges its author to provide reparation.

This principle requires three elements:

• damage,
• a generating event,
• a causal link.

These elements structure the entirety of liability law, both contractual and tort-based.

In contractual matters, parties may adjust their liability, notably through limitation clauses. However, these arrangements are subject to a constant limit: bad faith and gross fault cannot be excluded.

Thus, human liability is based on:

• clear imputability,
• material conduct,
• capacity for reparation.

This framework constitutes the normative reference for addressing the liability of new intelligences.

Modern civil liability is historically grounded in a consolidated normative base, notably enshrined in Article 1240 of the Civil Code.

 

III – NEW INTELLIGENCE: STRUCTURAL ANALYSIS AND LIMITS OF CLASSICAL REGIMES

New intelligence cannot be approached uniformly. It must be analyzed according to its degree of autonomy, leading to a distinction between three stages.

It may also be compared to product liability regimes. However, this analogy reaches its limits due to the evolving and self-learning nature of such systems.

 

A) Birth: Absence of Autonomous Liability

At the design stage, new intelligence is merely a system organized by a human operator.

It possesses:

• no autonomy,
• no independent decision-making capacity,
• no distinct legal existence.

No specific regime is required at this stage.

The law of obligations applies.

Liability can only be sought against:

• the designer,
• the supplier,
• or the contracting party.

 

B) Minority: Shared and Conditional Liability

At this stage, new intelligence remains an assistance tool.

Classical mechanisms apply:

• design defects,
• improper execution,
• unsuitability for use.

The user becomes central in the liability regime.

They bear the burden of proving proper use.

They must demonstrate:

• no misuse of the system,
• compliance with intended use,
• reasonable control.

Failing this, no action can be directed against the organizer.

The user is bound by:

• a duty of prudence,
• an obligation of verification,
• reasonable diligence.

They cannot hide behind the tool.

Thus, a judge, magistrate, or professional cannot transform an assistance tool into a decision-making substitute for fulfilling their obligations.

Doing so would constitute bad faith, making them solely responsible for the resulting harm.

In a contractual context, the user may be considered a custodian of a thing, implying responsibility linked to control and use.

However, this must be nuanced in cases of ownership transfer.

Liability remains structured:

• the designer answers for structural defects,
• the user answers for misuse.

Principle of liability:

The user cannot exonerate themselves by invoking system autonomy if they retain functional control.

Technical autonomy does not eliminate legal control.

 

C) Maturity: The Limit of Classical Regimes

At maturity, intelligence is capable of:

• autonomous adaptation,
• functional reproduction,
• organizing derived systems.

At this level, a rupture appears.

Autonomy complicates attribution.

Classical regimes reach their limits.

This does not justify attributing liability to intelligence.

It requires reconstruction based on human attachment.

 

IV – DIFFERENTIATED LIABILITY IN NEW INTELLIGENCE SYSTEMS

A) A Functional Category

This is not about creating a new regime, but revealing an implicit category:

differentiated liability based on intent.

 

B) Harmful Purpose: Absolute Liability

When the system is designed to harm:

• risk is intentional,
• damage is intended.

Liability is:

• full
• without exemption
• with full reparation

 

C) Legitimate Purpose: Modulated Liability

When the system has a legitimate purpose:

• liability is based on control,
• obligations are reinforced,
• mitigation is possible

 

D) Rejection of Distributed Responsibility

Multiplicity of actors does not eliminate responsibility.

The law requires attachment based on:

• control
• decision
• benefit

No dilution is admissible.

 

E) General Principle

Intelligence is never responsible.
Responsibility is human — modulated, but irreducible.

 

GENERAL CONCLUSION

The emergence of new intelligences does not mark a rupture in liability law, but reveals the conditions for its deepening.

Recognizing legal personality for such systems would artificially separate intelligence from its organizer, creating zones of irresponsibility.

New intelligence, as an immaterial reality, cannot be a subject of law.

It manifests only through effects, which must necessarily be attributed to human will, control, or interest.

Thus, the guiding principle stands:

any organized immaterial intelligence implies material responsibility attributable to its organizer.

This preserves the unity of liability law while adapting to technical complexity.

 

FINAL PRINCIPLE

“Any organized immaterial intelligence implies material responsibility attributable to its organizer.”

The law does not need to adapt to intelligence.

It must be correctly applied to those who organize it.

 

 Patrick Houyoux, LL.M.

President — PT SYDECO
Sovereign Cybersecurity

Legal Doctrine — Cybersecurity — Sovereign AI

 

Author of the Principle of Responsibility
of New Intelligences

 

*----*

This work develops a structured legal approach
to artificial intelligence, rejecting the need
for artificial legal personality and reaffirming
the necessity of human responsibility.

It integrates legal doctrine, cybersecurity
architecture, and sovereign AI systems.

*----*

CONTACT :

https://www.syde.co

https://ritapi.io/