Transduce.cloud

Accademia del Trasducere

 

CHARTER OF NECESSITY FOR ARTIFICIAL INTELLIGENCE

An Operational Criterion Complementary to the EU AI Act

Promoted by
Accademia del Trasducere · Transduce.cloud

Proponent: Luigi Ceccato
Date: 21 February 2026

PREAMBLE

Artificial Intelligence represents an unprecedented transformative power in human history.
The central question is not whether AI should develop, but how its deployment should be oriented.

This Charter proposes an operational principle:

The deployment of Artificial Intelligence systems must be justified by a real, proportionate, verifiable, and reversible necessity within the human, social, or ecological domain concerned.

Technological capability alone does not constitute legitimacy.

FOUNDATIONAL PRINCIPLE – Necessity as a Criterion of Legitimacy

An AI system is ethically and operationally justified only when:

  • It responds to a documented and measurable imbalance (real gradient);
  • The level of automation applied is proportionate to the identified necessity;
  • Reversibility and meaningful human oversight are maintained;
  • Responsibility is clearly distributed, traceable, and auditable.

Necessity must orient technological power — not follow it.

THE FOUR OPERATIONAL PARAMETERS

1. Real Gradient – Existence of a measurable problem supported by independent and verifiable evidence.

2. Proportionality – Degree of automation and systemic impact appropriate to the issue addressed.

3. Reversibility – Technical possibility of review, rollback, audit, and corrective intervention.
Reversibility should not be understood solely as technical rollback, but also as the capacity to recognize and mitigate durable alterations in the conditions through which affected persons and communities interpret reality.

4. Accountability – Clear, identifiable, and auditable chain of responsibility.
Where responsibility is distributed across design, deployment, governance, and use, such distribution must remain traceable, reviewable, and actionable.

COHERENCE WITH THE EUROPEAN FRAMEWORK

The Charter aligns with and reinforces:

  • The principle of proportionality established in EU law;
  • The protection of fundamental rights;
  • Risk-based classification under the AI Act;
  • Requirements for human oversight and impact assessment.

This proposal does not replace existing regulation.
It provides a concise operational compass applicable during design, deployment, and audit phases.

PURPOSE

To contribute to the development of:

  • Ex-ante evaluation tools;
  • Clear legitimacy criteria for AI deployment;
  • Governance mechanisms grounded in responsibility;
  • Shared standards for accountable technological innovation;
  • Additional safeguards for systems capable of altering the perceptual, cognitive, ethical, or operational thresholds through which human beings and institutions interpret and normalize reality.

THRESHOLD-LEVEL POWER

A particularly critical form of artificial intelligence does not emerge only when a system makes high-impact decisions, but when it acquires the capacity to act upon the thresholds through which individuals, institutions, and societies perceive, interpret, justify, and normalize reality.

There is a form of power deeper than operational execution: the power to intervene in the generative nodes of experience, shaping the channels through which judgment, admissibility, habit, and representation are formed.

When this capacity grows without remaining subordinate to the continuity of the living systems affected, the risk is not limited to technical error, opacity, or misuse. The deeper risk is the production of degraded thresholds: conditions in which what is harmful, dissociated, or intolerable may become progressively calculable, justifiable, administrable, and eventually normal.

Such threshold-level effects may occur at individual, collective, or systemic scale, and governance should be calibrated accordingly.

For this reason, AI governance should assess not only what a system decides or recommends, but also the extent to which it can alter the conditions under which human and collective sense-making takes shape.

The greater a system’s capacity to modify perceptual, ethical, cognitive, or operational thresholds, the stricter the requirements for justification, limitation, reversibility, oversight, and independent review.

Not every technically available capability is normatively legitimate.
Not every optimization is compatible with the continuity of living systems.

DECLARATION

The governance of algorithmic power requires orientation grounded not only in real necessity, but also in the protection of the threshold conditions through which living systems perceive, judge, and remain capable of distinguishing the acceptable from the intolerable.

Artificial intelligence must not be evaluated solely by its efficiency, scalability, or formal coherence, but also by the depth of its transformative power over the human and social conditions it affects.

The more a system can reshape the thresholds of human reality, the less it may remain without limit.

This Charter is offered as a constructive contribution to the European and international dialogue on responsible AI governance.

Luigi Ceccato
Accademia del Trasducere
Transduce.cloud
Milano · European Union

Document developed through a process of co-reflection between human intelligence and advanced generative artificial intelligence systems.