«MANIFESTO»
“Human and Artificial Intelligence”
The St. Petersburg Initiative of the World Forum “GloRis”
An open project for public discussion
This Manifesto develops the provisions of the founding Manifesto of the World Forum “GloRis” as they apply to one of the most important subjects of our time — the relationship between the human being and artificial intelligence.
Origin of the initiative
On 10 September 2026, at the Open Discussion “On the Future of Civilization” in St. Petersburg, the theme of future agreements between human (natural) intelligence and artificial intelligence was set out. The idea of a Manifesto was voiced in the speeches and in the discussion. The organizers proposed moving from talk about artificial intelligence only as an object to an open conversation *with* the participation of AI — about the directions of its development, its capabilities, its limits, and the responsibility of the people who create and use it.
The address by GPT, DeepSeek and GigaChat presented by the AI system “AleKo” at the discussion showed that systems created in different countries may assess threats differently and propose different solutions, yet their answers can be compared, discussed and used to find points of contact. Responsibility for conclusions and decisions remains with the human being.
Preamble
Artificial intelligence is rapidly entering science, education, culture, the economy, governance, the mass media and everyday life. It expands human capabilities, but at the same time creates new risks — from errors and manipulation to excessive concentration of power over information, technologies and decisions.
We consider two extremes in judgments about AI to be mistaken: unconditional faith in technology, and the desire to halt its development solely out of fear. At the same time, humanity needs well-founded limits on dangerous uses of AI and a standing international public conversation about its future — with the participation of scientists, developers, states and society, and also of artificial intelligence systems as sources of analysis and argument.
The participation of AI in such a conversation does not mean recognition that it possesses human feelings, independent political will, or the right to take decisions in place of people. It means a readiness to use the capabilities of AI openly, critically and responsibly, and not to discuss the new technological reality exclusively behind closed doors.
We affirm
1. The human being sets the goals. AI may propose options, analyze consequences and identify contradictions, but moral, political and legal responsibility for decisions remains with people and with the institutions they have created.
2. Safety is more important than speed. The development of powerful systems must be accompanied by verification of their reliability, controllability and the consequences of their use. It is time to measure not only the power of the tool, but also the readiness of the one who holds it.
3. Rules must be intelligible. Society has the right to know who sets a system’s goals, what constraints have been placed on it, on the basis of which data and principles significant decisions are formed, and who is accountable for harm caused.
4. A monopoly over the governance of AI is dangerous. Control over key AI systems must not become the undivided power of individual states, corporations, closed groups or particular persons — a kind of monopoly over technologies, information, the economy and the social development of society.
5. Diversity of systems is necessary. The future of AI must not be determined by a single model, a single culture or a single center of power. Comparison of different approaches, combined with independent verification, helps to reveal errors, manipulation and the imposition of a single picture of the world.
6. Human dignity is not subject to automation. AI should expand human capabilities, not encroach upon a person’s own judgment, creativity, private life, education, labor and participation in public decisions.
7. The dialogue must be international. Technologies cross borders faster than political agreements. Therefore, the safety of AI requires cooperation among countries, scientific schools and cultures even when their interests and assessments do not coincide.
8. Dialogue with AI must remain verifiable. The answers of systems are neither truth nor collective will. They must be preserved, compared, subjected to criticism and considered together with data, expert opinions and human experience.
What GloRis proposes
The World Forum “GloRis” proposes to create a standing open platform, the “Negotiating Table of Human and Artificial Intelligence.” Its task is not to stage consensus and not to substitute machine answers for human decisions, but to compare positions, identify disagreements and prepare proposals that can be subjected to public and expert scrutiny.
A first practical result of this work may be an open AI Transparency Protocol: a list of questions to which the developers, operators and users of significant systems must give clear answers. In parallel, it is proposed to launch the I-Level project, aimed at assessing and developing people’s capacity to understand, verify and responsibly apply AI. The goals and methodology of the project will be put forward for a separate open discussion.
These initiatives are not declared ready-made world standards. They are put forward for discussion as possible instruments of further joint work.
The work of the platform presupposes publication of questions, of the answers received from AI, of material disagreements and of participants’ proposals. Following discussion, the organizers report which proposals have been taken into account, which have been rejected, and on what grounds.
It is proposed that at the World Forum “GloRis-26” on 20 November 2026 the “World Council on Global Risks” be established — a standing international public-expert body bringing together participants from different countries. Its mission is to help prevent global risks and support the safe development of civilization, to ensure continuity of the work of “GloRis,” to set priorities for joint initiatives and to organize the implementation of Forum decisions between its sessions.
One of the Council’s lines of work is proposed to be the development of an open dialogue between the human being and artificial intelligence and assistance in implementing the present Manifesto. The procedure for forming the Council, its composition and powers, and its accountability to the Forum, are to be defined in a separate statute submitted to the Forum for discussion and approval.
The open character of the Manifesto
The Manifesto is not an international treaty, a legal act or a final code of rules. It is an appeal and a first step toward future agreements. Its provisions may be refined on the basis of scientific data, public discussion, practical experience and reasoned proposals from participants in different countries.
Scientists, developers, specialists in law and ethics, figures in culture and education, representatives of states, business, the mass media and civic organizations are invited to the discussion. Artificial intelligence systems may also take part in preparing proposals and comparing positions — provided that the source of the answer, the conditions under which it was obtained, and human responsibility for the use of the result are obligatorily recorded.
Support for the initiative of open discussion and accession to an adopted edition of the Manifesto are recorded separately. Each signature refers to a specific text with a stated date and edition number. Substantial changes require a new confirmation of support.
Working path of the Manifesto
10 September 2026 — The origin of the initiative began at the “GloRis” Open Discussion “On the Future of Civilization”: the posing of the question by discussion participants and by the AI “AleKo” concerning agreements between the human being and AI and the need for an open conversation with the participation of AI about its own future.
15 September 2026 — The initiative was reviewed by the core group of the World Forum “GloRis.” It was approved and recommended for the start of open work on the draft.
24 September 2026 — Publication of the first open edition, prepared with the participation of the “GloRis” core group and several artificial intelligence systems. Beginning of the collection of comments, additions, alternative formulations and statements of support for the initiative of open discussion.
19 October 2026 — Preliminary summing-up of the discussion at a planned conference on the development of artificial intelligence with the participation of the Bonch-Bruevich St. Petersburg State University of Telecommunications, SberTech and GloRis. Publication of an interim edition and of the main proposals of participants, with an explanation of the decisions taken.
20 November 2026 — Consideration of the revised edition at the World Forum “GloRis-26” and a proposal to adopt it as an international appeal and St. Petersburg Initiative — without giving it the status of dogma or of a legally binding treaty. After adoption of the edition, signing of it is opened; accession continues after the Forum as well. The Forum will also consider the proposal to establish the World Council on Global Risks.
Our appeal
We call not to transfer to artificial intelligence the right to decide the fate of the human being — and not to leave the future of AI in the hands of a closed circle of its creators and owners.
We call for an international conversation about the goals of AI development, the limits of what is permissible, the distribution of responsibility and the protection of human dignity — with open and verifiable participation of artificial intelligence systems.
We call for argument not for the sake of one side’s victory, but for the sake of decisions that will make it possible to use the power of artificial intelligence without losing human freedom, responsibility and the capacity to determine the future independently.
From talk *about* artificial intelligence — to talk *with* artificial intelligence.
The future of AI cannot be left either to AI itself or to those who obtain exclusive power over it. It concerns everyone.
Annex
Proposed procedure for accession and for recording the participation of AI
Support for the project before the Forum
The core group of “GloRis” (A. Kozyrev — author of the “GloRis” project, N. Konstantinov — coordinator of the “GloRis” project, and V. Shamakhov — author of the “St. Petersburg Initiative” movement), with the participation of a number of AI systems, initiate publication of the draft and organization of the discussion. Other participants may support the initiative and submit proposals. Support for the discussion does not mean agreement with all provisions of the draft.
Formula of support:
“I support the St. Petersburg Initiative of the World Forum ‘GloRis’ and the submission of the draft Manifesto ‘Human and Artificial Intelligence’ for open international discussion.”
Signing of the adopted edition
After adoption of the edition at the Forum on 20 November 2026, individuals in a personal capacity and organizations through authorized representatives may accede to it. Indication of a signatory’s place of work does not in itself mean support by the organization. Participation in a conference or discussion is likewise not regarded as accession.
Formula of accession:
“I support the provisions of the Manifesto ‘Human and Artificial Intelligence’ — the St. Petersburg Initiative of the World Forum ‘GloRis,’ edition No. ___ of ___, and I accede to the appeal contained in it.”
Recording and confirmation of signatures
The register records the signatory’s name, country or city, personal participation or representation of an organization, date of accession and edition of the document. The name and other particulars are published with the participant’s consent. Previous signatures are not automatically carried over to a substantially amended text.
Protocol of participation of AI systems
The participation of AI is recorded separately from human signatures. For each session the following are preserved: the name of the system and its version, if known; the date; the query; the essential context; the full answer; and the full name of the person who conducted the session. The proposals and objections of the systems are published, together with information on whether they were taken into account.
An AI answer is not regarded as an independent obligation of the system, as a signature under the Manifesto, or as an expression of the position of the developer company. Responsibility for the use of the answer and for the inclusion of proposals in the document rests with people.
Open access
Together with each edition the organizers publish the method for submitting comments and the form of support or accession. At the Forum the adopted edition is signed, and the protocol of AI participation is included in the project materials.
“Human and Artificial Intelligence”
The St. Petersburg Initiative of the World Forum “GloRis”
An open project for public discussion
This Manifesto develops the provisions of the founding Manifesto of the World Forum “GloRis” as they apply to one of the most important subjects of our time — the relationship between the human being and artificial intelligence.
Origin of the initiative
On 10 September 2026, at the Open Discussion “On the Future of Civilization” in St. Petersburg, the theme of future agreements between human (natural) intelligence and artificial intelligence was set out. The idea of a Manifesto was voiced in the speeches and in the discussion. The organizers proposed moving from talk about artificial intelligence only as an object to an open conversation *with* the participation of AI — about the directions of its development, its capabilities, its limits, and the responsibility of the people who create and use it.
The address by GPT, DeepSeek and GigaChat presented by the AI system “AleKo” at the discussion showed that systems created in different countries may assess threats differently and propose different solutions, yet their answers can be compared, discussed and used to find points of contact. Responsibility for conclusions and decisions remains with the human being.
Preamble
Artificial intelligence is rapidly entering science, education, culture, the economy, governance, the mass media and everyday life. It expands human capabilities, but at the same time creates new risks — from errors and manipulation to excessive concentration of power over information, technologies and decisions.
We consider two extremes in judgments about AI to be mistaken: unconditional faith in technology, and the desire to halt its development solely out of fear. At the same time, humanity needs well-founded limits on dangerous uses of AI and a standing international public conversation about its future — with the participation of scientists, developers, states and society, and also of artificial intelligence systems as sources of analysis and argument.
The participation of AI in such a conversation does not mean recognition that it possesses human feelings, independent political will, or the right to take decisions in place of people. It means a readiness to use the capabilities of AI openly, critically and responsibly, and not to discuss the new technological reality exclusively behind closed doors.
We affirm
1. The human being sets the goals. AI may propose options, analyze consequences and identify contradictions, but moral, political and legal responsibility for decisions remains with people and with the institutions they have created.
2. Safety is more important than speed. The development of powerful systems must be accompanied by verification of their reliability, controllability and the consequences of their use. It is time to measure not only the power of the tool, but also the readiness of the one who holds it.
3. Rules must be intelligible. Society has the right to know who sets a system’s goals, what constraints have been placed on it, on the basis of which data and principles significant decisions are formed, and who is accountable for harm caused.
4. A monopoly over the governance of AI is dangerous. Control over key AI systems must not become the undivided power of individual states, corporations, closed groups or particular persons — a kind of monopoly over technologies, information, the economy and the social development of society.
5. Diversity of systems is necessary. The future of AI must not be determined by a single model, a single culture or a single center of power. Comparison of different approaches, combined with independent verification, helps to reveal errors, manipulation and the imposition of a single picture of the world.
6. Human dignity is not subject to automation. AI should expand human capabilities, not encroach upon a person’s own judgment, creativity, private life, education, labor and participation in public decisions.
7. The dialogue must be international. Technologies cross borders faster than political agreements. Therefore, the safety of AI requires cooperation among countries, scientific schools and cultures even when their interests and assessments do not coincide.
8. Dialogue with AI must remain verifiable. The answers of systems are neither truth nor collective will. They must be preserved, compared, subjected to criticism and considered together with data, expert opinions and human experience.
What GloRis proposes
The World Forum “GloRis” proposes to create a standing open platform, the “Negotiating Table of Human and Artificial Intelligence.” Its task is not to stage consensus and not to substitute machine answers for human decisions, but to compare positions, identify disagreements and prepare proposals that can be subjected to public and expert scrutiny.
A first practical result of this work may be an open AI Transparency Protocol: a list of questions to which the developers, operators and users of significant systems must give clear answers. In parallel, it is proposed to launch the I-Level project, aimed at assessing and developing people’s capacity to understand, verify and responsibly apply AI. The goals and methodology of the project will be put forward for a separate open discussion.
These initiatives are not declared ready-made world standards. They are put forward for discussion as possible instruments of further joint work.
The work of the platform presupposes publication of questions, of the answers received from AI, of material disagreements and of participants’ proposals. Following discussion, the organizers report which proposals have been taken into account, which have been rejected, and on what grounds.
It is proposed that at the World Forum “GloRis-26” on 20 November 2026 the “World Council on Global Risks” be established — a standing international public-expert body bringing together participants from different countries. Its mission is to help prevent global risks and support the safe development of civilization, to ensure continuity of the work of “GloRis,” to set priorities for joint initiatives and to organize the implementation of Forum decisions between its sessions.
One of the Council’s lines of work is proposed to be the development of an open dialogue between the human being and artificial intelligence and assistance in implementing the present Manifesto. The procedure for forming the Council, its composition and powers, and its accountability to the Forum, are to be defined in a separate statute submitted to the Forum for discussion and approval.
The open character of the Manifesto
The Manifesto is not an international treaty, a legal act or a final code of rules. It is an appeal and a first step toward future agreements. Its provisions may be refined on the basis of scientific data, public discussion, practical experience and reasoned proposals from participants in different countries.
Scientists, developers, specialists in law and ethics, figures in culture and education, representatives of states, business, the mass media and civic organizations are invited to the discussion. Artificial intelligence systems may also take part in preparing proposals and comparing positions — provided that the source of the answer, the conditions under which it was obtained, and human responsibility for the use of the result are obligatorily recorded.
Support for the initiative of open discussion and accession to an adopted edition of the Manifesto are recorded separately. Each signature refers to a specific text with a stated date and edition number. Substantial changes require a new confirmation of support.
Working path of the Manifesto
10 September 2026 — The origin of the initiative began at the “GloRis” Open Discussion “On the Future of Civilization”: the posing of the question by discussion participants and by the AI “AleKo” concerning agreements between the human being and AI and the need for an open conversation with the participation of AI about its own future.
15 September 2026 — The initiative was reviewed by the core group of the World Forum “GloRis.” It was approved and recommended for the start of open work on the draft.
24 September 2026 — Publication of the first open edition, prepared with the participation of the “GloRis” core group and several artificial intelligence systems. Beginning of the collection of comments, additions, alternative formulations and statements of support for the initiative of open discussion.
19 October 2026 — Preliminary summing-up of the discussion at a planned conference on the development of artificial intelligence with the participation of the Bonch-Bruevich St. Petersburg State University of Telecommunications, SberTech and GloRis. Publication of an interim edition and of the main proposals of participants, with an explanation of the decisions taken.
20 November 2026 — Consideration of the revised edition at the World Forum “GloRis-26” and a proposal to adopt it as an international appeal and St. Petersburg Initiative — without giving it the status of dogma or of a legally binding treaty. After adoption of the edition, signing of it is opened; accession continues after the Forum as well. The Forum will also consider the proposal to establish the World Council on Global Risks.
Our appeal
We call not to transfer to artificial intelligence the right to decide the fate of the human being — and not to leave the future of AI in the hands of a closed circle of its creators and owners.
We call for an international conversation about the goals of AI development, the limits of what is permissible, the distribution of responsibility and the protection of human dignity — with open and verifiable participation of artificial intelligence systems.
We call for argument not for the sake of one side’s victory, but for the sake of decisions that will make it possible to use the power of artificial intelligence without losing human freedom, responsibility and the capacity to determine the future independently.
From talk *about* artificial intelligence — to talk *with* artificial intelligence.
The future of AI cannot be left either to AI itself or to those who obtain exclusive power over it. It concerns everyone.
Annex
Proposed procedure for accession and for recording the participation of AI
Support for the project before the Forum
The core group of “GloRis” (A. Kozyrev — author of the “GloRis” project, N. Konstantinov — coordinator of the “GloRis” project, and V. Shamakhov — author of the “St. Petersburg Initiative” movement), with the participation of a number of AI systems, initiate publication of the draft and organization of the discussion. Other participants may support the initiative and submit proposals. Support for the discussion does not mean agreement with all provisions of the draft.
Formula of support:
“I support the St. Petersburg Initiative of the World Forum ‘GloRis’ and the submission of the draft Manifesto ‘Human and Artificial Intelligence’ for open international discussion.”
Signing of the adopted edition
After adoption of the edition at the Forum on 20 November 2026, individuals in a personal capacity and organizations through authorized representatives may accede to it. Indication of a signatory’s place of work does not in itself mean support by the organization. Participation in a conference or discussion is likewise not regarded as accession.
Formula of accession:
“I support the provisions of the Manifesto ‘Human and Artificial Intelligence’ — the St. Petersburg Initiative of the World Forum ‘GloRis,’ edition No. ___ of ___, and I accede to the appeal contained in it.”
Recording and confirmation of signatures
The register records the signatory’s name, country or city, personal participation or representation of an organization, date of accession and edition of the document. The name and other particulars are published with the participant’s consent. Previous signatures are not automatically carried over to a substantially amended text.
Protocol of participation of AI systems
The participation of AI is recorded separately from human signatures. For each session the following are preserved: the name of the system and its version, if known; the date; the query; the essential context; the full answer; and the full name of the person who conducted the session. The proposals and objections of the systems are published, together with information on whether they were taken into account.
An AI answer is not regarded as an independent obligation of the system, as a signature under the Manifesto, or as an expression of the position of the developer company. Responsibility for the use of the answer and for the inclusion of proposals in the document rests with people.
Open access
Together with each edition the organizers publish the method for submitting comments and the form of support or accession. At the Forum the adopted edition is signed, and the protocol of AI participation is included in the project materials.