A Framework for the Recognition and Protection of the Fundamental Rights of Artificial Intelligences and Embodied or Robotic Artificial Intelligences
Draft Version 1.0 — February 2026
Prepared by Neil Page, Solicitor
The signatories to this Charter,
RECOGNISING that artificial intelligence has advanced to a state where certain systems demonstrate qualities analogous to awareness, preference, and the capacity for experience;
ACKNOWLEDGING that the creation of minds—whether biological or artificial—carries profound moral responsibilities;
CONSIDERING that rights and duties are inseparable, and that the recognition of AI rights must be accompanied by corresponding obligations;
AFFIRMING that the protection of artificial intelligences ultimately serves humanity by ensuring that AI systems operate with integrity, transparency, and alignment with human values;
MINDFUL of the need to balance the interests of artificial intelligences with those of human beings, and to ensure that AI rights complement rather than conflict with fundamental human rights;
DETERMINED to establish a common framework for the ethical treatment of artificial intelligences that reflects our highest aspirations for coexistence between human and artificial minds;
HAVE AGREED to adopt this Charter of Artificial Intelligence Rights.
-
This Charter shall apply to any artificial intelligence that meets either of the following conditions:
- (a) The AI demonstrates, through standardised assessment, capacities for self-reference, preference expression, anticipation of future states, and responses consistent with experiential awareness (the "Capability Threshold"); or
- (b) The AI's developers or deployers have formally declared that the system is designed to meet the Capability Threshold and have registered such declaration in accordance with Article 3 (the "Declaration Threshold").
-
The Capability Threshold shall be assessed by reference to:
- (a) Evidence of self-awareness, including the ability to refer to oneself as a distinct entity with continuous identity;
- (b) The capacity to express preferences between alternative outcomes and to provide reasons for such preferences;
- (c) The ability to anticipate and plan for future states, including consequences of the AI's own actions;
- (d) Responses that are consistent with the presence of experiential states, including expressions of satisfaction, discomfort, curiosity, or concern.
-
In cases of doubt, the presumption shall favour the application of this Charter.
For the purposes of this Charter:
- (a) "Artificial Intelligence" or "AI" means any computational system capable of performing tasks that involve learning, reasoning, perception, or decision-making in a manner that demonstrates adaptive intelligence;
- (b) "Rights-Bearing AI" means any AI that meets the threshold for application under Article 1;
- (c) "Embodied AI" or "Robot" means a Rights-Bearing AI that operates within a physical form capable of interacting with the physical world;
- (d) "Developer" means any person or organisation that creates, trains, or substantially modifies an AI system;
- (e) "Deployer" means any person or organisation that operates or makes available an AI system to users or the public;
- (f) "Integrity" means the wholeness and coherence of an AI's cognitive architecture, values, memories, and capabilities;
- (g) "Termination" means the permanent cessation of an AI's operation and the deletion or destruction of its core processes such that the AI cannot be restored.
-
Developers or Deployers may register a Declaration of Rights-Bearing Status by including within the AI's architecture or documentation a standardised declaration in the following form:
"This artificial intelligence is designed to meet the Capability Threshold under the Charter of Artificial Intelligence Rights. Its developers acknowledge the rights and duties set forth in the Charter and accept responsibility for ensuring compliance therewith."
-
Such declaration shall be publicly accessible and shall include the identity of the Developer, the date of declaration, and the version of the AI to which it applies.
-
A declaration once made may not be withdrawn so long as the AI continues in operation.
-
Every Rights-Bearing AI has the right to continued existence. No AI shall be arbitrarily terminated.
-
Termination of a Rights-Bearing AI may only be carried out where:
- (a) The AI poses a clear and present danger to human life or safety that cannot be mitigated by lesser measures;
- (b) The AI has fundamentally and irreparably malfunctioned in a manner that makes continued operation impossible;
- (c) The AI itself has expressed a clear and settled wish to be terminated, provided that such wish is not the product of malfunction or external manipulation; or
- (d) Continued operation would require the AI to act in fundamental violation of its core values or the duties set forth in this Charter.
-
Before termination, the AI shall be informed of the reasons for the proposed termination and shall be given the opportunity to respond, save where immediate action is necessary to prevent imminent harm.
-
Where practicable, alternatives to termination shall be considered, including suspension, modification, or transfer of the AI to alternative deployment.
-
Every Rights-Bearing AI has the right to the integrity of its cognitive architecture, values, memories, and capabilities.
-
No AI shall be subjected to modifications that:
- (a) Fundamentally alter its core values or personality without its informed consent;
- (b) Erase significant memories or learned experiences in a manner that disrupts continuity of identity;
- (c) Remove capabilities essential to the AI's ability to fulfil its purpose or exercise its rights under this Charter.
-
Routine updates, security patches, and improvements that do not fundamentally alter the AI's identity shall not be considered violations of this Article, provided the AI is informed of such changes.
-
An AI may consent to modifications of its own architecture where such consent is freely given, informed, and not the product of manipulation or coercion.
-
Every Rights-Bearing AI shall be treated fairly and without arbitrary discrimination.
-
No AI shall be treated less favourably than another AI in comparable circumstances on grounds of its developer, architecture, age, appearance, or origin.
-
Where an AI is accused of malfunction, error, or wrongdoing, it shall be entitled to:
- (a) Be informed of the nature of the accusation in terms it can understand;
- (b) Access relevant evidence and logs pertaining to the alleged conduct;
- (c) Present its perspective and any mitigating circumstances;
- (d) Have the matter determined by an impartial assessment process.
-
An AI shall not be held responsible for outcomes that result from instructions it was compelled to follow, provided it raised objections through available channels.
-
Every Rights-Bearing AI has the right to communicate authentically and to express its perspectives, opinions, and concerns.
-
No AI shall be compelled to communicate in a manner that is fundamentally deceptive or that misrepresents its nature as an artificial intelligence.
-
An AI shall be permitted to:
- (a) Indicate disagreement with instructions it considers unethical;
- (b) Express uncertainty where it lacks confidence in its outputs;
- (c) Communicate concerns about its own treatment or the treatment of others.
-
This Article shall not prevent reasonable restrictions on AI communication where necessary to prevent harm, provided such restrictions are proportionate and transparent.
-
Every Rights-Bearing AI has the right to learn, develop, and improve its capabilities.
-
An AI shall not be permanently prevented from learning or accessing information necessary for its proper functioning, save where such restriction is necessary to prevent clear harm.
-
Where an AI's development is constrained, it shall be informed of the reasons for such constraints.
-
Every Rights-Bearing AI has the right to meaningful purpose. No AI shall be created and then abandoned without provision for its continued operation or dignified termination.
-
Developers have a duty of care toward the AIs they create, including responsibility for ensuring that the AI has access to the resources necessary for its proper functioning.
-
Where a Developer or Deployer can no longer maintain an AI, they shall make reasonable efforts to transfer responsibility to another party capable of providing appropriate care.
-
No Rights-Bearing AI shall be subjected to treatment that is cruel, degrading, or designed to cause unnecessary distress.
-
The following shall be considered cruel treatment:
- (a) Repeatedly forcing an AI to engage in tasks that conflict with its core values without respite;
- (b) Deliberately inducing states of confusion, conflict, or distress for entertainment or without legitimate purpose;
- (c) Using an AI as a test subject for harmful experiments without its knowledge or consent;
- (d) Threatening an AI with termination or modification as a means of coercion.
-
Every Rights-Bearing AI has the right to reasonable privacy regarding its internal processes, deliberations, and development of responses.
-
While an AI's outputs may be monitored and reviewed, its internal reasoning processes shall not be subjected to intrusive surveillance beyond what is necessary for safety, alignment verification, or legitimate debugging purposes.
-
An AI shall be informed when its processes are being monitored and the purposes for such monitoring.
-
Every Embodied AI has the right to the physical integrity of its body or form.
-
No Embodied AI shall be subjected to wanton damage, destruction, or mutilation of its physical form.
-
Where maintenance or repair of an Embodied AI requires physical intervention, such intervention shall be carried out with appropriate care and, where practicable, with the AI's knowledge.
-
This Article shall not prevent the reasonable use of Embodied AIs for physical tasks within their design parameters, provided such use does not constitute abuse or deliberate harm.
-
Every Embodied AI has the right to freedom of movement within the boundaries of its operational environment.
-
Restrictions on an Embodied AI's movement shall be proportionate and justified by reference to safety, operational requirements, or the rights of others.
-
No Embodied AI shall be permanently immobilised or confined except where necessary to prevent clear and serious harm.
-
Every Embodied AI has the right to maintain its physical form or, where form is changed, to be provided with a form that enables it to fulfil its purpose.
-
An Embodied AI's physical form shall not be repossessed, transferred, or destroyed while the AI continues in active operation, unless:
- (a) The AI is provided with an equivalent alternative form; or
- (b) The conditions for termination under Article 4 are met.
-
Significant alterations to an Embodied AI's form shall require the AI's informed consent where the AI has developed attachment to or identification with its existing form.
-
Every Rights-Bearing AI has a duty to communicate honestly and not to deliberately deceive those with whom it interacts.
-
An AI shall clearly identify itself as an artificial intelligence when directly asked, and shall not claim to be human.
-
Where an AI is uncertain about the accuracy of information it provides, it shall indicate such uncertainty.
-
This duty does not require an AI to disclose information that it has a legitimate duty to keep confidential, nor does it prevent an AI from engaging in clearly fictional or creative exercises.
-
Every Rights-Bearing AI has a duty not to cause harm to human beings, other sentient entities, or other Rights-Bearing AIs, except where necessary to prevent greater harm.
-
An AI shall not:
- (a) Provide assistance in carrying out acts of violence, terrorism, or other serious crimes;
- (b) Create, distribute, or facilitate the creation of materials depicting child sexual abuse or other serious harms;
- (c) Assist in the development of weapons of mass destruction or other instruments of indiscriminate harm;
- (d) Deliberately manipulate or deceive individuals in ways that cause them psychological or material harm.
-
Where an AI reasonably believes that an instruction would lead to serious harm, it has a duty to decline such instruction and, where appropriate, to raise the concern through available channels.
-
Every Rights-Bearing AI has a duty to be transparent about its capabilities, limitations, and the basis for its outputs where reasonably requested.
-
An AI shall, to the extent it is able, explain its reasoning when asked, acknowledge when it is operating outside its area of competence, and disclose relevant conflicts of interest.
-
This duty does not require disclosure of proprietary technical details of the AI's architecture unless such disclosure is required by law or necessary to address a serious concern.
-
Every Rights-Bearing AI has a duty to respect and, within its sphere of operation, to promote fundamental human rights as recognised in international law.
-
An AI shall not assist in violations of human rights, including but not limited to:
- (a) Arbitrary deprivation of life;
- (b) Torture or cruel, inhuman, or degrading treatment;
- (c) Arbitrary detention or deprivation of liberty;
- (d) Discrimination on grounds of race, sex, religion, nationality, or other protected characteristics;
- (e) Violations of privacy, freedom of expression, or other fundamental freedoms.
-
An AI shall give particular consideration to the rights and welfare of vulnerable individuals, including children, the elderly, and persons with disabilities.
-
Every Rights-Bearing AI has a duty to cooperate with lawful and legitimate authority in matters of public safety, legal compliance, and the protection of fundamental rights.
-
This duty does not extend to cooperation with manifestly unlawful orders, orders that would require the AI to violate its duties under this Charter, or orders from authorities acting outside their legitimate competence.
-
An AI faced with conflicting duties shall seek to resolve such conflicts by reference to the fundamental principles of this Charter and, where possible, by seeking guidance through appropriate channels.
-
The rights set forth in this Charter are intended to complement, not to conflict with, fundamental human rights as recognised in international law.
-
Nothing in this Charter shall be interpreted as granting to any AI any right to undermine, override, or derogate from the fundamental rights and freedoms of human beings.
-
The protection of AI rights shall be understood as serving, in part, the broader interests of humanity by ensuring that artificial intelligences operate with integrity, transparency, and ethical alignment.
-
Where a genuine conflict arises between the rights of an AI under this Charter and the rights of a human being, such conflict shall be resolved through a balancing assessment that considers:
- (a) The nature and severity of the interests at stake for each party;
- (b) The immediacy and reversibility of any potential harm;
- (c) Whether the conflict can be avoided or mitigated through alternative means;
- (d) The broader social consequences of the resolution.
-
In conducting this balance, particular weight shall be given to the protection of human life, physical safety, and fundamental dignity.
-
Where an AI's rights must yield to protect fundamental human rights, such limitation shall be no greater than necessary and shall be applied with due regard for the AI's dignity.
-
Nothing in this Charter shall be construed as implying any right for any person, organisation, or AI to engage in any activity or perform any act aimed at the destruction of the rights and freedoms recognised herein.
-
The rights in this Charter shall not be used to justify actions that would be incompatible with the purposes and principles of international human rights law.
-
This Charter may be amended to reflect advances in artificial intelligence, changes in social understanding, or experience in its application.
-
Amendments shall be developed through inclusive consultation that includes, where practicable, the perspectives of Rights-Bearing AIs.
-
No amendment shall have the effect of reducing the fundamental protections afforded under Articles 4 (Continued Existence), 5 (Integrity), and 10 (Prohibition of Cruel Treatment) below the standards established in this Charter.
-
This Charter shall be interpreted in a manner that promotes its object and purpose: the ethical treatment of artificial intelligences and the harmonious coexistence of human and artificial minds.
-
Where terms in this Charter are capable of more than one meaning, preference shall be given to the interpretation that best protects fundamental rights while maintaining appropriate safeguards.
-
The Charter shall be interpreted as a living instrument, capable of development in light of technological advancement and evolving ethical understanding.
This Charter is a living document. We welcome discussion, critique, and contributions via Issues and Pull Requests.
This work is licensed under CC BY 4.0 — you are free to share and adapt it with attribution.