OMN hub cee

Building artificial intelligence without the people it affects is not a correctable omission: it is the origin of the problem. While disability continues to be treated as an exception in regulatory frameworks, society is condemned to legislate permanently from the margin, generating layers of correction over an architecture that was born incomplete. The AI Act (Regulation 2024/1689) does not incorporate disability as an autonomous category of protection at the design stage. It treats it as a reactive limit. A regulation that acts after the damage is not governance: it is consequence management. The first outcome that would make this Dialogue successful is the incorporation of disability as a structural axis at the origin of the design of all high-risk AI systems, with binding and verifiable obligations. Designing from the exception perpetuates the exception. No technical correction is possible over an architecture that excludes by construction. The second outcome is epistemic and without precedent in any international governance forum. Neurodivergent people possess transversal thinking that identifies systemic impacts invisible to linear thinking until the damage is irreversible. Their mandatory presence at 50% in all design, audit and governance teams of high-risk AI systems is not a demographic measure, nor does it respond to categories of sex or origin. It is a technical requirement measured exclusively by capacity, knowledge and contribution. Without transversal thinking at the origin, the architecture is structurally blind. The third outcome is legal. The AI Act sanctions the company economically. Article 9 of the ECHR protects the freedom of thought and conscience of the person. Between both exists a gap of impunity: when a system deliberately excludes or manipulates persons with disabilities, no individual bears criminal responsibility. As in patent law, where the violation of rights generates direct and non-transferable personal liability, cognitive manipulation through AI demands individual criminal consequences. Corporate sanctions do not repair a damaged conscience, do not restore a violated identity, do not return what an algorithm decided to silence. We can build our ai constitution since exception. 300 words.

• Protection and promotion of human rights
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Safe, secure and trustworthy AI

Disability remains structurally absent from the design origin of high-risk AI systems. The AI Act (Regulation 2024/1689) does not incorporate disability as an autonomous protection category in Articles 5, 10 or 27. It treats it reactively, after damage is produced. This is not a regulatory gap: it is a political choice that condemns persons with disabilities to inhabit a technological world designed without them, perpetually legislating from the exception. Human rights protection and the social and ethical implications of AI are therefore the most urgent priorities, because they expose the foundational contradiction of current governance: systems that formally prohibit discrimination while architecturally producing it. Transparency and accountability are inseparable from this contradiction. When an AI system excludes or manipulates a person with disability, the AI Act sanctions the company economically. Article 9 of the ECHR protects freedom of thought and conscience of the person. Between both instruments exists a gap of structural impunity: no individual bears criminal responsibility. As in patent law, where rights violations generate direct personal liability, cognitive manipulation through AI demands individual criminal consequences. Without that accountability, transparency is declarative, not operative. Safe and trustworthy AI cannot be achieved without resolving the epistemic deficit at the origin of design. Neurodivergent people possess transversal thinking that identifies systemic impacts invisible to linear cognition until damage is irreversible. A mandatory 50% inclusion of neurodivergent profiles in all design, audit and governance teams of high-risk AI systems, measured exclusively by capacity and contribution and independent of sex or origin, is not a diversity measure. It is a technical requirement. An architecture designed without transversal thinking is structurally blind to the consequences it produces. These four priorities are not parallel: they form a single argument. Rights without accountability are declarations. Accountability without epistemic diversity is correction after damage. Safety without inclusive design from the origin is consequence management. 300 words.

There is a category absent from all the thematic frameworks listed and from virtually all international AI governance forums: cognitive accessibility as an autonomous dimension of fundamental rights. Cognitive accessibility is not a subcategory of physical accessibility or digital inclusion. It is the condition that determines whether a person can form, express and exercise their thinking in environments mediated by AI systems. When those systems are designed exclusively from linear cognitive profiles, they do not only produce inaccessible interfaces: they produce architectures that predetermine which forms of reasoning, communication and existence are valid. This is not a technical problem. It is a violation of Article 9 of the European Convention on Human Rights, which protects freedom of thought and conscience, and of Article 9 of the United Nations Convention on the Rights of Persons with Disabilities, which establishes accessibility as a precondition for the effective exercise of all other rights. None of the themes listed by Resolution 79/325 names this dimension. Human rights protection presupposes it but does not operationalise it. Transparency and accountability cannot be exercised if the accountability system itself is cognitively inaccessible. Safety and trustworthiness of AI cannot be guaranteed if the teams evaluating it lack transversal thinking to detect its systemic consequences. The emerging issue that the Dialogue must incorporate is therefore twofold: the binding legal recognition of cognitive accessibility as an autonomous protection category in all AI governance instruments, and the obligation for persons with cognitive disability to participate in the design, audit and governance of the systems that affect them, not as beneficiaries but as epistemic authority. We need return our human ethics values fulfilling and protect human right 299 words.

The European Union has operated under the AI Act since 2024 with the declared ambition of leading AI governance grounded in fundamental rights. However, in the specific sector where the intersection between technology and disability is most visible — technology-focused special employment centres — governance gaps produce direct and immediate consequences that no current regulatory framework captures with precision. The most significant challenge is structural. Persons with disabilities working in technological environments interact daily with AI systems designed without them: selection tools, training platforms, performance evaluation systems and digital assistants that penalise interaction rhythms, forms of expression and non-linear cognitive structures. The AI Act does not require these systems to be audited from the perspective of cognitive accessibility. The European Accessibility Act addresses physical and sensory barriers but not cognitive ones. The result is an accumulation of invisible discrimination that is technically legal and structurally systematic. The opportunity is equally concrete. Spain has a legal framework for special employment centres that integrates technology and labour inclusion of persons with disabilities as an operational model, not as a declaration of principles. This model demonstrates empirically that cognitive diversity in technological teams does not reduce productive capacity: it expands it, because transversal thinking identifies systemic consequences that linear thinking does not detect. That operational evidence, combined with published forensic research on profiling bias in generative systems, constitutes a contribution that no other sector can offer to global AI governance. The gap between what the European framework promises and what it produces in this sector is not an anomaly: it is the most precise demonstration of why cognitive accessibility must become an autonomous governance category, binding and verifiable from the origin.

The AI Dialogue has an opportunity that no national or regional regulatory instrument can have: creating the space where frameworks that today operate in parallel — the European AI Act, national disability legislation, international human rights conventions — are articulated as a coherent system rather than coexisting as disconnected layers of partial obligations. Today there is an unresolved structural contradiction. Article 9 of the ECHR protects freedom of thought and conscience. The United Nations Convention on the Rights of Persons with Disabilities establishes accessibility as a precondition for the exercise of all other rights. The AI Act regulates the systems that violate both instruments. However, disability does not appear in the AI Act as an autonomous protection category at the design stage: not in Article 5, not in Article 10, not in Article 27. It is treated as a reactive limit, not as a constitutive principle. The result is that a system can be fully compliant with the AI Act and simultaneously exclude persons with disabilities in a systematic manner, because compliance is measured without reference to their experience. The Dialogue can close that gap through three operational commitments. First, a binding coordination mechanism between national AI Act supervisory bodies and CRPD monitoring committees, so that cognitive exclusion detected in a system automatically activates the rights protocol for persons with disabilities. Second, an international standard of cognitive accessibility in the design of high-risk AI systems, with legal standing equivalent to physical accessibility standards. Third, an internationally auditable register of cognitive diversity impact assessments, publicly accessible and verifiable. International cooperation on AI governance is not a diplomatic problem. It is a problem of legal architecture. The Dialogue has the authority to build it.

AI governance produces exactly what its architecture allows it to produce. And what its architecture allows it to produce is determined by who designs it, who audits it, and who decides what counts as harm. This is the central thesis of El silencio algorítmico de las smart cities europeas: that silence is not an accidental absence but a structural consequence of systems built from the exception. Systems that do not design for everyone from the origin, but correct for some from the margin. The AI Act is not European in its impact. It is global in its consequences. Any high-risk AI system operating under its framework — or operating outside it precisely because that framework does not reach it — affects persons with disabilities on every continent. And on every continent, disability continues to be treated as a category of exception: something added afterwards, corrected when damage has already occurred, mentioned in annexes but not in constitutive principles. A regulation built from the exception does not produce inclusion. It produces management of exclusion. Algorithmic silence is not a European metaphor. It is the global condition of any person whose way of thinking, communicating or existing was not considered at the moment when the architectural decisions that determine their access to rights, services and participation were taken. That moment did not happen in Brussels. It happened in every design room, every audit team, every governance forum that constituted its expert groups without the people whose lives their decisions would irreversibly affect. Geneva can be the first forum to break that chain. Or it can reproduce it with better declarations. 299 words.

The most absent voices in global AI governance forums are not those of associations or federations. Those organisations have consolidated access, funding and institutional representation. Their presence in international forums is frequent. Their epistemic independence, however, is limited: they operate within power structures that condition what they can say, before whom and in what terms. The biases they produce are not individual. They are organisational, and therefore invisible to those who exercise them. The truly absent voices are those of scientists and researchers who belong to the communities they study, who have built verifiable evidence on algorithmic bias from their own experience, and who simultaneously lead companies in sectors of real inclusion. This figure — the scientist-entrepreneur who inhabits the silenced community they investigate — has no recognised category in any international participation mechanism. They represent no one but their own knowledge. And precisely for that reason their voice is structurally distinct: it is not conditioned by representation mandates, institutional agendas or funding dependencies that limit what they can assert. Their inclusion in the Dialogue requires a mechanism of active identification based on three criteria exclusively: verifiable and published scientific knowledge, demonstrated operational experience in sectors that integrate technology and inclusion, and direct membership of the communities affected by the systems to be governed. Without intermediaries. Without umbrella organisations that filter or translate their position. With real authority at decision-making tables, not in symbolic opening panels. AI governance needs those who know the problem from the inside and have the scientific tools to name it with precision. Geneva must build the mechanism to find them. People how me

The most innovative formats are not the most technologically sophisticated. They are those that break the participation architecture that reproduces exclusion. The problem with international forums is not a lack of creativity in session design: it is that their formats are built for those who already know how to participate in them from an office, without having risked their own capital, without having managed teams of persons with disabilities in real technological environments, without having published verifiable forensic evidence on the systems they intend to govern. The most effective format for the Geneva Dialogue would be binding scientific-entrepreneurial testimony: sessions where researchers who are simultaneously entrepreneurs in real inclusion sectors present evidence built from operational practice and academic rigour, not from theoretical offices. Entrepreneurs who have invested their own capital to demonstrate that cognitive diversity in technological teams is not an ethical concession but a verifiable competitive advantage. That combination — personal economic risk, published scientific evidence and direct operational experience — produces knowledge that no institutional consultant can replicate. The second format is co-design from the origin: working groups constituted with 50% of persons with disabilities and neurodivergent profiles who have built companies, published research and managed real operations, selected exclusively by demonstrated capacity and contribution. Not reacting to already drafted proposals but participating in their drafting from the first word. The third format is public audit in session: high-risk AI systems evaluated in real time by teams that include scientist-entrepreneurs from affected communities, with documented and publicly accessible consequences. The Dialogue that wants to a big change how AI is built must listen to those who are already doing it with their own resources.

The most concrete and least cited model in international AI governance forums was born from an experience that no public policy document records: the systematic institutional abuse that pushes persons with disabilities out of the systems that should protect them. OMN Hub, based in Spain, was not created from a market opportunity. It is a response built from the consciousness of having suffered what it seeks to correct. That is its deepest legitimacy and its most irreplicable contribution to any debate on AI governance. Eg: once institutional company have a dualism and never will promote the change because it's a millionaire business This origin matters because it defines the type of knowledge it produces. Not knowledge about exclusion observed from outside. It is knowledge of exclusion inhabited, scientifically documented and transformed into an operational model with private capital and measurable results. The interactions between neurodivergent persons and generative AI systems, documented through forensic methodology with SHA-256 dual hashing, demonstrate that profiling bias does not reside in the data but in the architecture of the systems themselves. That evidence has a value that no laboratory disconnected from practice can produce. OMN Hub operates at the intersection where the governance of AI is not a theoretical debate but a daily operational reality: teams of persons with disabilities working in technology, generating verifiable scientific evidence from their own experience, and demonstrating that cognitive inclusion is not an ethical concession but a technically superior architecture. Every day of operation is data. Every interaction documented is evidence. Every result achieved with private capital is proof that what governance forums declare as aspiration is already being built by those the system attempted to silence. The concrete practice that this model contributes to global governance is ethical certification of high-risk AI systems audited by scientist-entrepreneurs who belong to affected communities, who have published verifiable evidence and who have risked their own capital to demonstrate that the system was wrong. 300 words.