Generative AI now shapes what billions of people encounter online, filtering and producing the information they access daily. These systems are fundamentally tools of communication and knowledge distribution, making how the world regulates them a matter of profound consequence for what citizens can express, discover, and understand. The UN's Global Dialogue on Artificial Intelligence Governance, scheduled for July 6-7 in Geneva, offers a critical moment to embed freedom of expression at the heart of global AI governance.

A foundation worth building on

Recent years have seen meaningful momentum toward rights-based AI governance. In February 2025, the Paris AI Action Summit produced a declaration signed by 58 nations committing to anchor AI policy in human rights and sustainability. The absence of the US and UK from this accord signals a troubling divergence among major AI powers, underscoring why Geneva's outcome matters urgently.

This commitment gained reinforcement at the institutional level months later. In October 2025, the UN Special Rapporteur on freedom of expression joined counterparts from the Organization for Security and Co-operation in Europe (OSCE), the Organization of American States (OAS), and the African Commission on Human and Peoples' Rights (ACHPR) in issuing a joint declaration. The statement emphasized that freedom of expression must be woven throughout every stage of AI development, from initial design and training through to real-world deployment.

These expert-level positions built upon commitments UN Member States had already made. In 2024, governments reached consensus through the Global Digital Compact to safeguard freedom of expression and access to information while addressing potential harms, and to avoid information restrictions that violate international law. That same Compact established the upcoming Geneva Dialogue. Even the OECD AI Principles, revised in 2024, tied information integrity safeguards to respect for freedom of expression. These instruments create a baseline that Geneva can strengthen and expand.

The New Delhi Declaration: an opportunity not taken

India's AI Impact Summit this year represented a setback for human rights protections in AI governance. More than 90 countries and international organizations endorsed the New Delhi Declaration, which called for "industry-led voluntary measures" and framed AI's mission as advancing "social good"—language devoid of accountability mechanisms or enforceable rights standards.

Analysis of the summit's agenda revealed a striking imbalance: terms like "innovation," "growth," "efficiency," and "productivity" dominated high-level sessions, while "human rights," "accountability," and "surveillance" barely appeared in plenary discussions. Geneva's structure differs markedly. The Dialogue includes a dedicated human rights cluster and an independent scientific panel—safeguards the New Delhi framework deliberately omitted. This positioning allows Geneva to remedy what New Delhi left undone by restoring human rights to its proper place in AI governance.

Why freedom of expression is central to AI

Generative AI systems now serve over a billion users worldwide. In 2025, information retrieval emerged as the dominant use case for generative AI across six countries—Argentina, Denmark, France, Japan, the UK, and the US. Most internet users encounter AI-generated summaries regularly through Google AI Overviews or Bing Copilot Search, whether or not they directly interact with chatbots. This ubiquity makes AI governance inseparable from the governance of speech itself.

International human rights law offers a sturdy foundation. Article 19 of the International Covenant on Civil and Political Rights, the primary global instrument protecting freedom of expression, guarantees the right to "seek, receive and impart information and ideas […] regardless of frontiers" through any medium. This expansive language provides a solid basis for protecting AI-generated expression and information in the digital age.

Governments have repeatedly invoked information accuracy and public order to suppress online political speech, jeopardizing criticism of state authority, journalism, and civil society advocacy. While information integrity represents a genuine concern, applying it without Article 19's safeguards creates danger. The Article 19 framework requires that any restriction be established by law, serve a legitimate purpose, and be necessary to achieve that purpose—a tripartite test that balances harm prevention with democratic values. This approach enables policymakers to address real dangers like non-consensual intimate imagery and child sexual abuse material while preserving democratic discourse.

Why rights protections matter in practice

The past three years have demonstrated that protecting freedom of expression and access to information extends far beyond theoretical principle. Governments have repeatedly weaponized vague safety arguments to suppress AI outputs.

In India, when Google's Gemini generated responses in February 2024 characterizing the prime minister's policies in language some experts called fascist, authorities accused Google of violating IT Rules and demanded pre-clearance for all AI model releases. The signal to developers was unmistakable: government-displeasing outputs carry legal risk. India's three-hour removal rule for AI-generated content, enacted in early 2026, intensified this pressure. Platforms must delete flagged unlawful material within three hours of government notification, with no distinction between political satire, journalism, and genuinely harmful content. The rule affects political opponents, journalists, ordinary citizens, and civil society alike. Direct content blocking orders in India's digital censorship infrastructure surpassed 24,000 in 2025, and the AI removal rule extends that apparatus into a new domain. Yet India is not alone in weaponizing public order justifications.

Turkey blocked access to Grok content in July 2025 after the xAI chatbot generated statements deemed insulting to President Recep Tayyip Erdoğan. The government cited public order protection—precisely the rationale Article 19's proportionality requirements constrain.

These pressures extend beyond nations with established censorship records. In the US, domestic disputes over AI neutrality have complicated a principled international defense of freedom of expression. The European Union's AI Act and Digital Services Act contain systemic risk provisions expansive enough to function as speech restrictions, with national regulators already criticizing chatbots for generating "biased" political outputs—a pattern warranting close attention.

Most significantly for Geneva, China's influence is not confined to its borders. In July of last year, China released its Global AI Governance Action Plan, invoking "public good" and "safety" as organizing principles. China's domestic AI regime mandates adherence to "core socialist values" and prohibits outputs challenging state authority or social stability. This framework is being actively promoted in the multilateral forums that will shape whether freedom of expression receives protection in global AI governance.

What the Global Dialogue can achieve

The UNESCO Recommendation on the Ethics of AI from 2021 offers a model for rights-centered AI governance. It requires that any rights limitation satisfy the tripartite test of legality, legitimacy, and proportionality; prohibits AI systems for social scoring or mass surveillance; and mandates that AI actors respect rights throughout the AI lifecycle. This template deserves renewed attention. Building on this foundation, the Global Dialogue can advance further by establishing freedom of expression and access to information as operative standards and applying Article 19's tripartite test to AI content governance decisions.

The Dialogue's human rights cluster and Independent International Scientific Panel can operationalize commitments UN Member States have already made. The following steps warrant consideration:

  • The Co-Chairs' Summary and any working group mandates should explicitly name freedom of expression and access to information—mirroring the Global Digital Compact's language—with civil society and independent experts playing substantive roles.
  • The Scientific Panel should evaluate current AI-enabled surveillance and content authenticity capabilities to fully assess potential harms to freedom of expression alongside privacy concerns.
  • The Dialogue should confirm that Article 19's requirements of legality, legitimacy, and necessity remain central throughout AI governance discussions, particularly regarding information integrity measures and any content restrictions.

AI represents the information technology of coming decades, and its governance demands careful attention to rights protection. Geneva offers the venue where rights-respecting standards can become global norms, transcending national and regional boundaries.

Source: Tech Policy Press