Manifesto for a Sovereign AI

Manifeste IA Souveraine




1. Preamble:

We, the proponents of sovereign Artificial Intelligence (AI), recognize the enormous potential of AI to transform our society and improve the lives of individuals. However, we note that AI is currently centralized, controlled by a small number of global technology companies, and used primarily to maximize profits, without regard for the interests and needs of users and local communities. This leads to inequality, excessive dependency, potential risks to privacy, security and sovereignty of nations.
We strongly believe that in the public interest, we need to change the paradigm and promote a sovereign AI, developed locally, and deployed on a decentralized, relocalized and user-governed infrastructure. We believe in an AI that benefits everyone, encourages local innovation, and serves humans, while respecting ethical principles, human rights and cultural diversity.

2. Fundamental principles :

1. Sovereignty: We advocate the principle of digital sovereignty, so that every nation, community and individual has the right to control their own AI and decide how it is used in their local environment, in accordance with the laws
and cultural principles of their country.
2. Localization: We promote the local development of AI, encouraging the creation of research, development and training centers in the territories, in order to foster the emergence of local expertise and high quality jobs. We also encourage the use of local data to train AI models, which reflect cultural, social and economic specificities.
3. Participatory governance: We support AI governance based on active participation of users and stakeholders to ensure ethical, transparent and fair decisions about AI. We emphasize the need to create
local advisory councils, composed of users, experts and civil society representatives, to guide the development, deployment and use of AI.
4. Decentralization: We advocate a decentralized infrastructure for AI deployment, based on small, clustered interconnected infrastructures relocated at customer sites. This reduces dependence on large centralized data centers,
improve data security and privacy, maintain equitable and resilient access to AI at all points, including rural, peripheral, and developing areas. We oppose the concentration of data storage and processing capabilities among a few players and reject any monopoly or oligopoly situation in this area.
5. Ethics and responsibility: We emphasize ethical values in the development and use of AI. We defend the obligation on transparency of algorithms, protection of privacy, responsibility of AI developers in designing fair and unbiased models, and
fair and unbiased models, and in considering the social, environmental, and economic impacts of AI. We call for the adoption of ethical standards and best practices in the field of AI, as well as the establishment of mechanisms for
accountability of companies and institutions using AI.
6. Responsible use of AI: We call for ethical principles in the design, development and use of AI models, ensuring that they remain fair, impartial and transparent. We also encourage the establishment of
independent monitoring and evaluation mechanisms to ensure compliance with ethical and local standards. We promote the use of AI to solve, not enhance, societal challenges, such as health, education, the environment and the fight against
7. Training and Awareness: We believe that training and awareness are essential to promote sovereign AI. We encourage the training of local AI talent, with an emphasis on equal access. We also advocate for public awareness of AI issues, explaining the benefits, risks, and challenges associated with the technology, and encouraging open and informed debate about its use and impact in society.

3. Conclusion

We support the vision of sovereign, local AI, governed by those who use it, deployed on decentralized infrastructure and relocated close to users. We encourage accountability of AI developers, citizen participation in AI governance, and the creation of an ethical, fair, and transparent ecosystem.
By adopting this manifesto, we commit to promoting sovereign AI that respects local norms, protects human rights, fosters innovation, and contributes to the well-being of society at large. We call on governments, businesses, communities, researchers, and citizens to join this vision and collaborate for the implementation of AI as a beneficial, equitable, and humanity-serving tool.


Appendix: Why a decentralized and locally owned infrastructure for AI?

An infrastructure for AI that is decentralized and relocated as close to the users as possible has several key advantages:
1. Local Sovereignty: By taking a decentralized approach, each customer can exercise control over their data and the AI deployed in a controlled environment. This ensures that decisions about AI are made locally, taking into account local laws, regulations and cultural values.
and cultural values of each region or country. It also avoids dependency on large technology companies and foreign powers, and in turn preserves the digital sovereignty of nations.
2. Flexibility and performance: A decentralized infrastructure allows to adapt to the needs and preferences of each customer. This architecture is based on small storage and data processing units, micro-clouds. Each micro-cloud will be configured according to the specific requirements of the deployed AI, the local environment and the users’ needs. This also offers greater flexibility in the deployment of AI, more freedom to optimize performance. Moreover, in proximity, response times between machines are accelerated, and their maintenance improved by local and identified teams.
3. Confidentiality and data protection: AI processes large amounts of potentially sensitive data. A decentralized infrastructure reinforces the confidentiality of data by avoiding their massive transfer to centralized servers. The multiplication and interconnection of micro-clouds ensures redundancy and backup between units, while allowing the creation of private watertight spaces. This resilient network is a solution to the risk that centralization poses to large units in the event of a disaster. With data remaining
remains local to the customer’s premises, which reduces the risk of data breaches, information leaks and privacy breaches. In addition, the use of data in accordance with local laws and regulations is guaranteed.
4. Innovation and diversity: A decentralized infrastructure encourages innovation by allowing local developers to experiment, create and deploy their own AI solutions. By promoting diversity of approaches and models. Healthy competition between local developers is stimulated, which can lead to faster technological advances and more tailored solutions.
5. Transparency and accountability: In a decentralized infrastructure, AI decisions are more transparent and accountable. Customers have better control over the decisions made by the AI deployed in their environment, they maintain a better understanding of the decision processes, and this allows for better accountability in case of errors or biases. The trust of users and stakeholders in AI is fostered, which is essential for its adoption and acceptance in society.

In summary, by taking a decentralized approach to infrastructure, we can ensure that AI is used responsibly, ethically and transparently, while preserving digital sovereignty, and fostering innovation in AI.


Appendix: U.S. Data Collection and Infrastructure Control

The United States is a major player in the field of artificial intelligence (AI). As a result, there are concerns about U.S. control over global AI data collection and infrastructure. Some reasons for these concerns include:
1. Data collection: U.S.-based technology companies, such as Google, Facebook, Amazon, and others, amass huge amounts of data around the world. This data includes sensitive personal information, usage preferences, consumer habits… There are concerns about how this data is collected, stored, and used, including privacy, security, and potential surveillance.
2. Control of infrastructures: The United States has significant control over AI infrastructures, through servers, data centers and communication networks that support the deployment and management of AI systems. This control is global. There are many questions about the dependence on these infrastructures and the concentration of power in the hands of the United States.
3. Technological inequalities and inequitable access: The concentration of development and control of AI technologies in the United States may contribute to widening or deepening technological inequalities between countries and regions.
4. Extraterritoriality of U.S. laws: U.S. laws, such as the Clarifying Lawful Overseas Use of Data Act, may allow U.S. authorities to access data stored abroad under certain circumstances. This raises concerns about sovereignty, privacy, and protection of data of users or companies located outside the U.S., as the U.S. justifies through its legal arsenal a digital preemption capability.
5. Geopolitics and surveillance: As a superpower, the United States is involved in geopolitical, national security and surveillance issues. This raises questions about the training and potential use of AI for surveillance, intelligence gathering or other intrusive
intelligence gathering or other intrusive activities.
All of these concerns are not specific to the U.S., they can apply to other countries and technology players around the world. But because of its dominant position, the U.S. tops the list of state actors that could use AI at the expense of citizens or organizations.


Appendix: China’s data collection and desire to control AI infrastructure

China is currently the United States’ main competitor for global dominance, including as an actor in the strategic development of AI, and as a user of AI for surveillance and coercion. As such, China has developed its own position.
1. Digital Sovereignty and Data Protection: China seeks to exercise sovereign control over the data collected on its territory. It has adopted strict legal and technical regulations on data collection, storage, processing and protection.
protection of data. China considers that the collection of data from its territory is a matter of sovereignty and national security, and therefore favors local infrastructures.
2. Control of infrastructure and governance: China is putting in place controls on AI infrastructure operating in its territory. Foreign companies working in China are invited to sign partnerships with local companies and eventually share their technology with them.
their technology with them. This approach aims to ensure that governance and decisions on AI issues are carried out in accordance with Chinese laws and policies.
3. Geopolitical strategy and technological independence: China views AI development as a strategic issue and therefore focuses on technological independence. Policies promote the development of local AI, but also the collection of data
operated on Chinese territory, which ensures direct access to the information needed to train AI models and improve algorithms. Reducing China’s dependence on foreign AI technologies and strengthening its competitiveness in the global market are at the heart of this
market are at the heart of this approach.
4. Information control and national security: Data collection and control of AI infrastructure ensure national security and protect the interests of the state. Strict regulations govern the collection and management of data, to prevent sensitive or even strategic information from being shared.
sensitive or even strategic information from being shared or stored abroad. China claims to protect its citizens’ data, as well as the country’s economic and geopolitical interests. China states its desire to become the world leader in AI.
5. Risks of Surveillance and Privacy Violations: However, there are legitimate concerns about government surveillance or the use of data for censorship, surveillance and social control. These policies may infringe on
infringe on users’ fundamental rights; AI potentially amplifies this trend. China has an Internet censorship and surveillance system known as the Great Firewall of China, which blocks access to certain foreign websites and applications
deemed contrary to national interests.

In conclusion, China is taking a sovereign approach in response to, but also in place of, U.S. domination. Within its borders, it uses a technological arsenal that reinforces its authoritarian domination over the population through control and surveillance. AI is an existential issue for China.


Appendix: European position on data collection and control of AI infrastructures

The European Union (EU) takes a balanced approach and formally emphasizes the protection of individual rights, digital sovereignty, security, transparency and the promotion of innovation. France, a member of the EU, shares this approach while maintaining
some specificities.
1. Protection of individual rights and privacy: Regarding the collection and use of data by AI technologies, the EU insists on the protection of individual rights. The EU’s General Data Protection Regulation (GDPR), which came into force in 2018,
regulates the framework and enforcement.
2. Digital Sovereignty and Security: The EU theoretically advocates for digital sovereignty, or the ability for EU member states to control their own AI infrastructures and ensure their security. France has adopted a posture in favor of digital sovereignty, and
France has adopted a posture in favor of digital sovereignty, and is taking steps to promote the development of sovereign AI, through dedicated research centers, public-private partnerships, and investment schemes for start-ups.
3. Transparency and ethics: The EU supports transparency and ethics in the development and use of AI. It makes developers responsible from the design to the deployment and use of AI systems. France supports this approach through guidelines
guidelines, codes of conduct and certification mechanisms.
4. Promotion of innovation: The EU and France promote innovation in the field of AI, through the funding of research and the promotion of entrepreneurship in technology. The EU has set up programs such as Horizon 2020. The EU’s position differs from that of the US and China in several aspects. Where the U.S. takes an approach focused on massive data collection and the primacy of technology companies, the EU emphasizes the protection of individual rights, digital sovereignty, and the safety of AI infrastructure. China, on the other hand, follows the path of centralization and also advocates for massive data collection in contrast to the EU’s focus on privacy.


Appendix: Interference between the visions of the United States, China, and the EU on AI

AI is at the heart of the global competition between the United States, China, the European Union, and a few players such as Japan or Russia, each defending their own approach. The interference between these visions can impact the development, deployment and governance of AI.
1. US vision: Considered to be at the head of the race, the United States, with its large leading companies, dominates the sector. The US vision of AI emphasizes the massive collection of data to drive models, and is strongly oriented towards
innovation and economic competitiveness. As a result, companies like Google, Amazon and Facebook, have access to huge amounts of data from around the world, giving them a strategic advantage in the field.
However, this approach can be experienced as predatory by other nations, and raises questions about sovereignty, respect for privacy, intellectual property, freedoms; the concentration of this digital power in the hands of a few companies also questions the power of the state in the face of these actors. The United States faces challenges related to the regulation of AI, whose constantly evolving legal framework is struggling to keep up with the rapid pace of technological progress.
2. China’s vision: Competing with the United States, China sees AI as a strategic priority and aims to become a world leader in this field by 2030. The separation of the Chinese internet, and the creation of a sovereign AI motivates China’s current very “independent” position. China’s vision of AI also relies on massive data collection by the government and companies to train models. The state supports and promotes research in a very active and planned way. One of the biggest
fields of application is the surveillance of populations for social stability and homeland security purposes.
This Orwellian approach taken by China is causing reactions to regular human rights violations. In addition, China is also confronted with the quality and reliability of the data collected, and arouses suspicion about the protection of intellectual property.
of intellectual property, which can limit the competitiveness of its companies in international markets.
3. EU Vision: The EU is taking a more balanced approach to AI, emphasizing the protection of individual rights, digital sovereignty, security, transparency, and ethics. The EU has strict regulations in place to protect individual privacy and rights, such as the GDPR. The EU insists on international cooperation, collaboration between its member countries and respect for European cultural norms.
But this European approach is slower in terms of innovation and competitiveness compared to the US and China. European companies are confronted with problems of access to data and financing, which limits their technological
technological advancement. In addition, regulatory fragmentation within the EU can also create barriers to the adoption and deployment of AI on a European scale. To conclude, these weaknesses favor lobbying by foreign firms at the expense of local solutions, reinforcing
solutions, reinforcing the EU’s dependence on the United States. The term “digital colonization” is sometimes used to describe this situation.
4. Winning vision: It is difficult to say which vision will ultimately win in the AI race, as this will depend on many factors, including the ability to innovate, access data, meet ethical standards, and respond to societal needs.
However, a balanced approach that considers the economic, social, ethical, and regulatory challenges of AI, while protecting individual rights and promoting international cooperation, could be the key to sustainable and
responsible development of AI on a global scale.
In conclusion, the interference between the visions of the United States, China, and the European Union on AI are complex and lead to tensions on various aspects. The winning vision will ideally be the one that can reconcile technological innovation, protection of individual rights, respect for ethical norms, and international cooperation for responsible development of AI on a global scale. However, it faces the reality of an “imperial” struggle between China and the United States, while the EU is trying to continue to exist between these two dominant players, it would have to come up with a truly sovereign vision, and practice a form of protectionism to survive as a global player.


Appendix: Adoption Methodology for Promoting Sovereign AI

1. Awareness and Engagement: Awareness is key. Government, communities, businesses, researchers, civil society, or the general public should be introduced to the benefits of sovereign AI. Awareness campaigns, workshops, conferences or public
public discussions are all tools to propagate the principle of a sovereign AI. Themes would emphasize privacy, sovereignty, security, eco-responsibility, and ethics.
2. Multi-actor collaboration: The implementation of a sovereign AI requires the collaboration of different public, private and associative actors. Public-private partnerships, industrial alliances, academic collaborations and multi-stakeholder consultations must foster a collective approach, so that AI can be
collective approach, so that AI is not only an issue of economic competition, but rather a development tool in the general interest.
3. Local development of AI: in parallel to the constitution of digital infrastructures under local control, it is essential to facilitate the development of AI in proximity to users. A local and decentralized operation avoids the capture of data by dominant central actors in disregard of
central and dominant actors in disregard of the sovereignty of citizens. We will favor local actors for the training of researchers and engineers, by encouraging the creation of startups or local companies specialized in AI, by facilitating the collaboration between
companies and universities in the nearby territory to promote innovation.
4. Appropriate regulation: The formation of a specific regulatory framework is inevitable to guide the development and use of sovereign AI. The defense of technological sovereignty, data protection, transparency, accountability and ethics in
the use of AI must motivate the legislator. However, be careful not to hinder innovation.
5. Economic incentives: The government can put in place incentive policies, such as tax credits, grants, public investments and funding programs, to encourage companies and researchers to develop and adopt sovereign AI solutions.
sovereign AI solutions. It would be desirable to also activate local market protection measures to encourage the adoption of indigenous solutions.
6. Evaluation and continuous improvement: An iterative and feedback-based approach is needed. Authorities, businesses, and users must monitor the results and impacts of sovereign AI, with a view to adjusting and improving its performance, security, privacy, and impact on the environment.
security, privacy and impact on society. Particular importance will be given to feedback and user needs.

The promotion of sovereign AI should involve local authorities so that they equip themselves while respecting the perimeters that frame the protection of citizens’ data. This use of local AI would increase the performance of services in the territories for the benefit of their citizens, and would become a powerful tool for economic and social development.



scroll to top