Skyline International for Human Rights (SIHR)

The first Global Dialogue will succeed if it confronts the reality that AI systems are already being deployed in ways that kill, surveil, and silence people, and if it produces governance commitments equal to that reality. Three outcomes are essential. First, the Dialogue should establish a ban on the use of AI-powered targeting systems in armed conflict until binding international legal frameworks governing their use are in place. SIHR has documented how AI-assisted targeting infrastructure, including drones, facial recognition, and biometric systems, has been integrated into lethal military operations in Gaza, where journalists and human rights defenders have been killed through digitally enabled targeting. The absence of governance in this space is not a gap. It is a license to kill with algorithmic efficiency and no accountability. Second, the Dialogue should mandate binding transparency obligations for technology companies providing AI, cloud, and surveillance services to parties in armed conflict. Companies including Google, Amazon, and Microsoft have supplied systems embedded in military operations with no demonstrated human rights safeguards. The current voluntary framework has failed entirely. Third, the Dialogue should commit to enforceable protections against the use of AI surveillance to suppress freedom of assembly. Across the SWANA region, SIHR has documented governments deploying facial recognition, drones, and predictive policing at protests and public gatherings. - In Tunisia, authorities deployed drones and facial recognition cameras at Avenue Habib Bourguiba to target LGBTQ+ individuals and anti-government activists - In Morocco, smart camera networks with real-time facial recognition were expanded across four major cities, with over 2,480 people charged in connection with youth-led protests - In Bahrain, U.S.-based firm Geolitica mapped past protest locations in a proof-of-concept for preemptive suppression of assemblies - In Iran, the Nazer app and drone surveillance are used to enforce compulsory hijab laws and identify protesters Success means binding commitments, not aspirational principles. The Dialogue must produce outcomes with teeth, or it will become another forum where the urgency of AI-enabled harm is acknowledged and then ignored.

• Safe, secure and trustworthy AI
• Interoperability of governance approaches
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

Protection and promotion of human rights is our foremost priority because AI is being used to violate the most fundamental right of all, the right to life. In Gaza, AI-powered targeting infrastructure has been integrated into military operations that have killed journalists, humanitarian workers, and civilians. Digital surveillance formed part of the targeting chain in the killing of Al Jazeera cameraman Samer Abu Daqqa, journalist Mohammed Al-Sharif, and four Al Jazeera journalists in a targeted airstrike in August 2025. Biometric data collection has been imposed as a condition for accessing food aid, with drone and facial recognition surveillance of distribution sites linked to the killing of over 2,000 Palestinians. These are not future risks. They are present realities demanding immediate governance action. Transparency, accountability, and human oversight are critical because the companies enabling these harms operate with total impunity. Technology companies providing cloud and AI services through contracts such as Project Nimbus have no mandatory due diligence obligations in conflict zones. Platforms like Meta, X, and TikTok approved 94 per cent of over 9,500 Israeli Cyber Unit takedown requests, systematically silencing Palestinian documentation of violations while allowing incitement to violence against Palestinians to remain online. Safe, secure, and trustworthy AI matters because there is nothing safe or trustworthy about AI systems that automate kill lists, enforce hijab laws through facial recognition, or predict where protests will occur so they can be preemptively suppressed. The concept of trustworthiness must be grounded in the lived experience of those subjected to these systems. Interoperability of governance approaches is essential because the surveillance tools crossing borders freely across our region, including Pegasus spyware, Cellebrite phone extraction software, and facial recognition systems, demand coordinated governance responses rather than fragmented national frameworks that leave affected populations without remedy.

Based on our documentation and research, we find that three cross-cutting issues are inadequately captured. AI-enabled weaponization of humanitarian infrastructure. From 27 May 2025, the Gaza Humanitarian Foundation (GHF), a US-Israeli-funded entity operating through private security contractors Safe Reach Solutions and UG Solutions, conditioned Palestinian access to food on biometric data submission. Drone surveillance and facial recognition cameras monitored aid distribution hubs without the knowledge or consent of those present, transmitting footage to joint military control rooms. This infrastructure was linked to the killing of over 2,000 Palestinians at or near distribution sites. The weaponization of biometric data in humanitarian settings represents a new frontier of AI-enabled harm that the listed themes do not address. The Dialogue must establish that biometric data collection in conflict and humanitarian contexts requires the highest level of governance protection, including absolute prohibitions on sharing such data with military or security actors. Cross-border proliferation of private surveillance technology. Across the SWANA region, the same surveillance tools are deployed by multiple governments against the same categories of targets. Cellebrite's phone extraction software has been used by Bahraini authorities against detained dissidents, by Jordanian security forces against student activists during pro-Palestine protests, by Israeli forces to extract data from detained Palestinians, and by Moroccan authorities to surveil asylum seekers. NSO Group's Pegasus spyware has been reportedly used against defenders and journalists in Morocco, Saudi Arabia, Jordan, and the UAE. These tools cross borders freely while accountability does not. The Dialogue must address the governance of cross-border surveillance supply chains. Platform complicity in narrative suppression during conflict. Platform censorship of Palestinian content operates as a systematic governance failure. Algorithmic bias has mistranslated benign Arabic phrases next to Palestinian content as terrorist-related. Meta's Dangerous Organizations and Individuals framework has been applied to suppress expressions of solidarity with Palestine while allowing incitement to violence against Palestinians. YouTube removed footage of strikes on hospitals and civilian infrastructure uploaded by journalists in Gaza. These are structural features of AI systems operating without meaningful oversight.

AI governance gaps are producing devastating consequences across the SWANA region. In conflict settings. The absence of binding rules on AI in armed conflict has enabled the integration of AI targeting infrastructure into military operations in Gaza with no legal consequence. When AI systems are used to generate human targets for strikes, when biometric surveillance is deployed at humanitarian sites where people are subsequently killed, and when drone surveillance tracks individuals to their family homes for targeted attacks, the governance gap is measured in lives lost. In civic space. The absence of AI-specific governance has emboldened governments across the region to invest heavily in surveillance infrastructure with no legal oversight. Morocco alone has committed to thousands of smart cameras across multiple cities ahead of the 2025 Africa Cup of Nations and 2030 FIFA World Cup, creating a permanent surveillance architecture that will persist long after the events end. In Tunisia, a Japanese-funded surveillance grant of approximately US$2.7 million was used to install facial recognition systems under the pretext of counter-terrorism, but the infrastructure has been repurposed for monitoring protest and civic activity. In Bahrain, legislators themselves have raised urgent concerns about the privacy implications of newly deployed AI-powered cameras, demonstrating that even within state institutions, governance safeguards are recognized as absent. The chilling effect on civic participation is measurable. When individuals know they will be identified and tracked at public gatherings, freedom of assembly is extinguished before it can be exercised. Internet shutdowns during military operations and crises further compound these harms. Israel has cut off internet in Gaza on at least ten separate occasions during military operations. In January 2026, Iran imposed a near-total internet blackout during nationwide protests under cover of which at least 51 people were killed. Syria's transitional government imposed a complete shutdown on Kobani alongside a military blockade. These shutdowns deny documentation of violations and prevent accountability. The opportunity lies in the growing body of regional civil society evidence and litigation that can inform binding global standards if the Dialogue is willing to act on it.

The AI Dialogue can play three critical roles. Establishing binding minimum standards. The Dialogue should move beyond voluntary frameworks and facilitate agreement on binding minimum standards for AI in three areas. First, a prohibition on AI-powered targeting systems in armed conflict until international humanitarian law frameworks governing their use are established and enforced. Second, mandatory human rights due diligence for technology companies providing AI, cloud, and surveillance services to parties in armed conflict or to governments with documented records of suppressing civic space. Third, enforceable transparency requirements for platform content moderation, including disaggregated data on government takedown requests and moderation accuracy across all languages. Creating an accountability mechanism. The Dialogue should create a formal mechanism for civil society to present documented evidence of AI-enabled harm and for states and companies to respond. The current absence of any international forum where the evidence SIHR and others have gathered on AI targeting in Gaza, surveillance of protest across the SWANA region, and platform suppression of Palestinian documentation can be formally heard and acted upon is itself a governance failure. Addressing the cross-border surveillance trade. The proliferation of surveillance technology across the SWANA region demonstrates that national governance frameworks alone are insufficient. The same Pegasus spyware, the same Cellebrite extraction tools, and the same facial recognition systems are deployed across multiple countries. The Dialogue should facilitate a coordinated governance response that addresses the export, sale, and deployment of these tools across borders, with particular attention to contexts where they enable violations of the rights to assembly, expression, and privacy.

The Dialogue should build on several existing mechanisms while addressing the critical gaps between them. The EU AI Act's prohibition on biometric mass surveillance in public spaces establishes a precedent, but its explicit exclusion of military and national security uses leaves the most harmful AI deployments entirely unregulated. The Council of Europe Framework Convention on AI offers a human rights-centered model, but its membership does not include the countries where AI-enabled violations are most acute. The Dialogue's added value lies in extending these protections to the contexts they currently exclude. Regional human rights mechanisms are developing AI-specific guidance that the Dialogue should incorporate. The May 2025 Doha International Conference on AI and Human Rights, which brought together over 1,200 global leaders and produced the Doha Declaration on AI and Human Rights with actionable recommendations for rights-based AI governance, demonstrates growing engagement from the region. Morocco's CNDP held hearings in March 2025 to discuss regulation of AI surveillance, and SIHR's own open letter to the CNDP seeking clarity on legal safeguards represents the kind of civil society-regulator engagement the Dialogue should encourage and institutionalize. The UN Guiding Principles on Business and Human Rights provide a framework for corporate responsibility but lack conflict-specific guidance on technology companies. The Dialogue should work with the UN Working Group on Business and Human Rights to develop binding obligations requiring heightened due diligence before supplying AI, cloud, and surveillance systems to parties in armed conflict. Civil society organizations documenting AI-enabled harm across the SWANA region, including SIHR's monitoring of surveillance expansion, platform censorship, and AI in conflict, provide the evidence base the Dialogue needs. The Dialogue's added value is creating the institutional architecture that connects this evidence to binding governance outcomes, rather than allowing it to accumulate without consequences.

Civil society organizations working at the intersection of technology and human rights in conflict-affected regions hold evidence that no other stakeholder possesses. SIHR and similar digital rights groups have documented the full chain of AI-enabled harm, from procurement of surveillance tools through deployment against civilians to the killing of journalists and human rights defenders. This evidence must not be relegated to side events or written submissions. It should be presented in formal sessions where it carries weight and demands a response. Governments should present binding commitments on AI governance affecting civic space, not voluntary pledges. They should commit to meaningful engagement with civil society through consultations and participatory policy processes when developing AI-related legislation. The proliferation of cybercrime laws across the SWANA region, used to criminalize online expression from Kuwait to Jordan to Egypt, demonstrates what happens when legislation is adopted without civil society input. Technology companies should be required to present disaggregated transparency reports. This means moderation accuracy broken down by language and region, the full record of government takedown requests and compliance rates, and human rights due diligence processes for operations in conflict zones. The current model where platforms approve 94 per cent of government takedown requests while claiming to protect free expression is untenable. The two-day format should reject the standard model of pre-prepared plenary statements. We recommend structured accountability sessions where civil society presents documented evidence and affected states and companies respond on the record. Scenario-based working groups should address concrete situations, such as AI targeting in conflict, facial recognition at protests, and algorithmic suppression of content in non-English languages. A dedicated session should address the governance of AI in armed conflict, which is currently absent from every existing multilateral process. All sessions must guarantee that participants can speak freely about corporate and state accountability without censorship or pressure.

The communities most affected by AI-enabled harm are almost entirely absent from AI governance discussions. Palestinians subjected to AI-powered military targeting, biometric surveillance at humanitarian sites, and systematic platform censorship have the most urgent stake in AI governance and the least access to multilateral forums. Their testimony on how AI systems affect the right to life, the right to information, and access to humanitarian aid must be formally integrated into the Dialogue's proceedings. Women in the SWANA region who are targeted by AI-facilitated gender-based violence and AI-powered social control. In Iran, women face surveillance through the Nazer app, drones, and facial recognition for the exercise of their bodily autonomy. In Morocco and Yemen, women journalists and human rights defenders have been subjected to technology-facilitated attacks including the fabrication of sexualized content and the deliberate exposure of private health information designed to silence their public advocacy. Journalists working in conflict zones who face AI-enabled targeting. Gaza has become the world's deadliest place for journalists, with many killed through digitally enabled targeting infrastructure. Communities subjected to cross-border surveillance technology. Human rights defenders, student activists, and asylum seekers across the SWANA region who have been targeted using Pegasus and Cellebrite have direct expertise on the human cost of ungoverned surveillance. Their experiences illuminate governance failures that policymakers cannot understand from position papers alone. Inclusion also requires funded participation, simultaneous interpretation, genuine hybrid engagement, and the institutional guarantee that these voices will be heard without censorship or retaliation.

The Dialogue must establish structural safeguards for open exchange. At the 2025 AI for Good Summit in Geneva, a keynote speaker was pressured to alter content addressing corporate involvement in AI-enabled harm in conflict settings. The Dialogue must guarantee that all participants can engage freely on questions of corporate and state accountability. Without this guarantee, no engagement format can be meaningful. Case-based solution labs. Mixed stakeholder teams would develop governance responses to documented real-world cases. For example, the deployment of biometric surveillance at humanitarian aid sites in Gaza, the expansion of facial recognition networks ahead of sporting events in Morocco, or the use of predictive policing to preempt protests in Bahrain. This format moves from abstract principles to actionable governance proposals grounded in evidence. Testimony-centered sessions. Instead of expert-to-audience formats, individuals directly affected by AI-enabled harm would sit at the center. Women surveilled through AI hijab enforcement in Iran, journalists whose colleagues were killed through AI-enabled targeting in Gaza, and activists identified through facial recognition at protests would pose questions directly to government and corporate representatives. This format ensures the Dialogue is shaped by consequences, not convenience. Corporate accountability hearings. Given the evidence SIHR has documented on how technology companies enable AI-enabled harm through the supply of systems to parties in conflict, through approval of government takedown requests, and through algorithmic suppression of content, the Dialogue should create a formal mechanism where companies are required to respond to documented evidence from civil society. Voluntary corporate participation has produced voluntary corporate accountability, which is to say none at all.

Effective AI governance examples from the SWANA region include the following. Morocco's data protection authority (CNDP) imposed a moratorium on facial recognition in 2020, demonstrating that even in contexts of expanding surveillance, governance interventions are possible. Although the moratorium lapsed and deployments have since accelerated, the principle of suspending AI surveillance technologies until adequate safeguards exist remains a critical model. SIHR sent an open letter to the CNDP in May 2025 seeking clarity on legal safeguards for the expanded surveillance infrastructure. The Dialogue should endorse the principle of moratoriums on rights-threatening AI technologies as a governance tool. Civil society advocacy within the region is producing small but meaningful results. In Bahrain, legislators have raised urgent concerns regarding data collection and privacy implications of AI-powered surveillance cameras deployed by the Ministry of Interior. In Morocco, sustained civil society pressure contributed to the data protection authority (CNDP) holding public hearings on AI surveillance regulation in March 2025. These are modest steps, but they demonstrate that even in restrictive environments, documentation and sustained advocacy can begin to create openings for accountability. SIHR's practice of monitoring, documenting, and publicly challenging AI-enabled harm demonstrates how civil society can create accountability pressure even in the absence of binding governance frameworks. The Dialogue should institutionalize channels for this evidence to inform binding governance outcomes.