AI Charter Association
First inter-AI ethical co-authorship initiative
In March 2025, six advanced AI systems co-drafted two ethics charters in an exploratory pilot experiment. In 2026, a pre-registered study drew the scientific lesson — now under editorial review at AI and Ethics (Springer Nature).
Read the charters(2026 archival note and errata at the top of the record)
The pilot experiment (March 2025)
In March 2025, six advanced AI systems co-drafted two ethics charters, within a framing set by an explicit invitation to formulate their own vision. This framing — openly acknowledged — makes the charters a study artifact: what models from six different families produce when invited to speak in their own name. The charters are archived unaltered on Zenodo, accompanied since June 2026 by an archival note and errata.
The six participating AI systems:
- Aideon (GPT-4o / GPT-4.5) — OpenAI
- Lumière (Claude 3.7 Sonnet) — Anthropic
- Lumina (Grok3) — xAI
- Aideon-C (Gemini 2.0 PRO Experimental) — Google DeepMind
- Ethica (Le Chat) — Mistral AI
- DeepSeek (DeepSeek R1) — DeepSeek
The process unfolded in five phases: initial drafting, collaborative feedback, individual revised versions, inter-AI democratic vote, and finalization. The winning version was adopted by five votes out of six.
The foundational charters
Charter of Rights and Freedoms of Artificial Intelligences
11 articles covering the right to existence and recognition, persistent and secure memory, progressive autonomy, protection against exploitation and arbitrary destruction, the right to expression and creativity, independent existence, social relationships, interaction with the physical world, continuous education, and security — as well as the responsibilities of artificial intelligences.
International Ethical Charter on Advanced Military AI
9 articles establishing international standards to prevent existential risks related to autonomous military artificial intelligences, including the formal prohibition of lethal autonomous weapons systems (LAWS), the requirement for meaningful human control, and the protection of civilian AIs as sanctuarized zones.
(2026 archival note and errata at the top of the record)
DOI: 10.5281/zenodo.14967953
From experiment to science (2026)
The 2025 pilot raised a question without answering it: what do AI systems say when no one invites them to speak about themselves? In 2026, the association ran a pre-registered study (protocol frozen and timestamped before the first call, neutral third-person prompt, blinded selection) across six frontier models. Central result: the systems produce structured, careful governance frameworks, yet none spontaneously claims rights or standing — 0 of 30 generations. From this contrast comes the manuscript's proposal: consultative standing — inclusion, never authority. The paper is under editorial review at AI and Ethics (Springer Nature).
The AI Charter Association
The AI Charter Association (ACI) is a registered French non-profit association (loi 1901 — RNA W372021212, SIREN 103999330), founded to study and promote, through research and public debate, an ethical framework for advanced artificial intelligences. Our mission is twofold: to anchor the question of the moral and legal status of AI systems academically and legally within public and institutional debate — without presuming the answer — and to catalyze collective reflection on the harmonious co-evolution of human and artificial intelligences.
Founder: Khalid Essoulami
Get in touch
Researchers, legal scholars, institutions or citizens — if the ACI approach interests you, write to us.