Freedom Online Coalition (FOC)

- A true, multistakeholder forum and process, in compliance with the NETmundial+10 Multistakeholder Statement; - An ability of non-government actors to participate, speak and interact with government actors and other stakeholders; - An outcome document that reflects the views and priorities of all sectors of society, and that creates an equal playing field for follow up and future Dialogues; - Agreement on the development and convergence of international AI standards, including standards for safety, transparency, evaluation, and risk classification across the AI lifecycle (design, development, deployment, and monitoring). These standards should be interoperable and inclusive of perspectives from the Global South.

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

In line with TFAIR's mandate, our interest focuses on how AI affects human rights, fundamental freedoms and democratic processes: - Without embedding transparency, accountability, and human oversight and human rights safeguards throughout the lifecycle of AI systems, AI will continue to be misused systematically to suppress dissent, manipulate public discourse, and enable unlawful and arbitrary digital surveillance. - Even without abuse, AI systems have an inherent tendency towards discrimination which may reinforce inequalities and amplify gender-based violence. Without appropriate guardrails for safety, security and trustworthiness, these risks also reduce the acceptance of AI within societies.

#NAME?

An example of the misuse of AI to manipulate public discourse is its effects on free and pluralistic information ecosystems, including independent journalism, which are essential to democratic participation. The same is true of the work of human rights defenders. AI systems can undermine the foundations of our shared information environment in multiple ways, particularly during elections, without proper safeguards. Current governance and transparency mechanisms are insufficient to evaluate how AI systems prevent and mitigate discourse manipulation, especially in underrepresented linguistic and cultural contexts where datasets are scarce or insufficiently labelled, limiting effective oversight and accountability. Among those most at risk of AI-related harms are women and girls in all their diversity. These harms include lack of access and representation in training data, and discrimination through algorithmic bias. They are compounded by the exclusion of diverse linguistic and cultural knowledge in AI training, raising concerns about representation and challenging the assumption that more data leads to equity. The impacts of AI are not evenly distributed. In many contexts, especially in the Global South, communities are affected by data extraction, outsourced data labour, and environmental pressures linked to digital infrastructure, while having limited influence over how systems are designed or regulated. This raises concerns about participation, accountability, and fairness in global AI governance. Furthermore, AI systems facilitate gender-based violence, including online harassment and image-based abuse such as non-consensual intimate imagery, as well as surveillance, disinformation, and other forms of repression. These practices undermine democratic governance, public trust, and gender equality. A key governance gap is the lack of widely adopted, enforceable AI standards. Without shared benchmarks and auditing frameworks, it remains difficult to assess compliance and ensure accountability, particularly in contexts with limited capacity. The environmental impact of AI is not fully known, despite growing demands on energy, water, and land.

As only UN-led global conversation, the Dialogue can: - Set governance priorities, including policy recommendations; - Be a model for multistakeholder processes; - Engage with Scientific Panel to research efficacy and practicalities of proposed policy recommendations; - Provide an exchange of experiences and best practices for mutual learning; - Ignite and increase global conversations on accountability frameworks relating to AI governance; - Facilitate international coordination on AI standard setting, including alignment between existing initiatives ( ISO, IETC , ETSI etc.) and ensuring that developing countries can meaningfully contribute to and adopt these standards.

TFAIR would like to point to itself as a partner for the AI Dialogue that can offer its substantial expertise for the thematic session on human rights. TFAIR has a diverse membership comprising 17 FOC member governments, 15 organisations from the FOC Advisory Network (FOC-AN), and 30 external stakeholders, including other experts in the field and additional government representatives. As pointed out in the co-chairs' draft note on themes and structure, this session could be co-chaired by a Member State and a relevant stakeholder. As TFAIR's membership comprises both states and civil society organisations, our members could take over both of these roles. We would also be happy to provide scene-setting interventions and suggest experts for panel discussions e.g. from the FOC Advisory Network.

In order to come to a true exchange of experience, stakeholders of various backgrounds should have a chance to present their experience in the format of impulse statements or panels, to be followed by interactive Q&A sessions without need for prior inscription. Breakout sessions with a smaller number of participants may allow better interaction. When it comes to panels, these should be steered by their moderator so that panelists react to each other's interventions. Panelists should not start with prepared statements, and there should be no more than 3 or 4 panelists in any panel. Furthermore, the consultations so far have shown a very substantial interest in the Dialogue from a broad range of stakeholders, including civil society organisations. All of these should get a chance to voice their views at the Dialogue. Consequently, enough time needs to be allocated for this. The Co-Chairs' draft note foresees mainly plenary sessions, with only 4 hours allocated to breakout sessions with at most two sessions in parallel. While the EU understands the rationale of this approach, we are afraid that it just will not produce enough opportunities for all stakeholders to meaningfully participate in the discussions. It is therefore recommended to move away from plenary sessions and allow more parallel sessions, possibly also more than two sessions at the same time.

In most discussions, governments and big enterprises have an advantage in terms of visibility and resource input. In order not to disadvantage civil society organisations and small enterprises further, do not put speakers in plenary sessions in any order of importance, but assign slots randomly, mixing the different categories of stakeholders. It is TFAIR's experience that such a mix makes discussions most fruitful.

We believe that meaningful engagement results from different stakeholder groups discussing common issues. In addition to focusing discussions around thematic issues - as the Co-Chairs are planning to do - it would be beneficial to connect government representatives with stakeholders from their countries. For example, space could be given so that representatives from the Government of Ghana meet directly with Ghanaian academics, industry representatives and civil society. These types of direct conversations create opportunities for the most pressing issues at the national level to be discussed by relevant experts, which will lead to better governance of AI domestically and internationally.

TFAIR has formulated Joint Statements on Artificial Intelligence and Human Rights and on Responsible Government Practices for AI Technologies that were adopted by the FOC membership and serve as potential guidelines for AI use and regulation and also as examples of what multi-stakeholder groups can agree on on these topics. The statements contain actionable recommendations and can inform the thematic session on human rights and AI at the dialogue.