Amnesty International

Amnesty International is a London-based international human-rights organisation founded in 1961 and operating across more than 150 countries; its AI-and-surveillance work — concentrated inside its Amnesty Tech programme — constitutes one of the most resourced civil-society operations in the make-AI-good field. The organisation's technology work spans biometric surveillance and facial-recognition bans, a business-model critique of surveillance capitalism, autonomous-weapons prohibition, predictive-policing documentation in the UK, a rights-based challenge to the data pipelines underpinning commercial generative AI systems (May 2026), and — from mid-2026 — a military-AI accountability campaign and a landmark country-level investigation of AI-driven surveillance in Argentina. Amnesty International's inclusion in this corpus rests on its demonstrated record of grassroots-organising engagement: the Ban the Scan campaign builds community-level coalitions with local civil-society organisations in each target jurisdiction; the global #BanBS letter was co-drafted as a 175+-organisation mobilisation vehicle; and the Algorithmic Accountability Toolkit is designed to transfer investigative capacity to CSOs, impacted communities, and local activists. These grassroots-engagement layers distinguish the organisation's AI work from the institutional-policy-only operations the corpus leaves out.

Founding and structure

Amnesty International was founded in London in May 1961 by British lawyer Peter Benenson following a newspaper article calling for an international campaign on behalf of political prisoners. It is now one of the world's largest human-rights organisations, with national sections, satellite offices, and grassroots activist networks across more than 150 countries and territories. The London-based International Secretariat coordinates the global network, with Amnesty Tech as the specialist programme for technology-and-human-rights work.

Amnesty Tech and the Algorithmic Accountability Lab

Amnesty Tech is the organisation's global technology and human rights programme, structured across four focus areas: spyware and cyberattacks targeting activists and journalists; surveillance and state use of AI and automation; big tech and social media accountability; and children and young people's rights in digital environments. Within Amnesty Tech, the Algorithmic Accountability Lab specialises in research, campaigning, and advocacy on automated state systems. In December 2025, Amnesty Tech launched its Algorithmic Accountability Toolkit — a comprehensive guide for civil-society organisations, journalists, impacted people, and community organisations investigating AI-facilitated harms in public-sector welfare, policing, healthcare, and education systems. The toolkit was framed explicitly as a capacity-transfer instrument: the Lab's own investigative methodology and framework handed outward to civil society actors who lack the resources to build it from scratch. Damini Satija (Director (interim) of Amnesty Tech by 2026, who founded the Algorithmic Accountability Lab) and Matt Mahmoudi (AI and human rights researcher) are the two named public-facing researchers behind the programme's recent output.

Ban the Scan — the facial recognition campaign

The organisation's most sustained grassroots-oriented AI campaign is Ban the Scan, launched on 26 January 2021 with New York City as its first target. The launch framed facial recognition as a technology that "violates the right to privacy and threatens the rights to freedom of peaceful assembly and expression" and — critically — as a system that amplifies racist policing by disproportionately misidentifying Black and other minoritised people. The Ban the Scan campaign hub carries the tagline "Your face is being tracked. Find out where. Help ban facial recognition technology." and has operated as both an online campaign tool and a grassroots organising vehicle. The New York launch ran in coalition with AI for the People, the Surveillance Technologies Oversight Project, the Immigrant Defense Project, the New York Civil Liberties Union, the New York City Public Advocate's office, The Privacy NY Coalition, and State Senator Brad Hoylman; it deployed a public-facing crowdsourced mapping tool through Amnesty Decoders — inviting members of the public to contribute to a geolocated database of New York's facial-recognition-capable surveillance camera infrastructure. Amnesty International's February 2022 research publication drew on the map data to demonstrate that NYPD facial-recognition-compatible camera density is highest in communities historically over-targeted by stop-and-frisk, reinforcing the campaign's central claim about racism and policing.

Ban the Scan subsequently extended to two further phases. In November 2021, Amnesty International carried the campaign into India in a three-organisation partnership with the Internet Freedom Foundation and ARTICLE 19, targeting the Hyderabad Command and Control Centre — the Telangana state surveillance system capable of integrating data from up to 600,000 cameras simultaneously — under the #BanTheScan framing that IFF and Project Panoptic had already built into the Indian civil-society register. In 2023, Amnesty extended the campaign to the Occupied Palestinian Territories, documenting Israeli authorities' deployment of the "Red Wolf" facial recognition system to track Palestinian movement in the West Bank and East Jerusalem. The three-jurisdiction arc of the campaign — New York, Hyderabad, Palestine — is the corpus's clearest cross-regional instance of a single civil-society framing travelling from a US city-level campaign into international human-rights accountability terrain.

The global #BanBS coalition and biometric surveillance

On 7 June 2021, Amnesty International was one of six co-drafting organisations behind the global open letter calling for a ban on biometric surveillance technologies — the letter that launched the #BanBS coalition in coordination with Access Now, EDRi, Human Rights Watch, the Internet Freedom Foundation, and the Instituto Brasileiro de Defesa do Consumidor (IDEC). The letter collected 175+ signatures from civil-society organisations, activists, researchers, and technologists across 55 countries at launch; it named a categorical prohibition demand — prohibiting the use of facial recognition and remote biometric recognition in publicly accessible spaces by both public bodies and private entities where such use could enable mass surveillance or discriminatory targeted surveillance — and framed the prohibition as three specific technology bans: remote facial recognition in publicly accessible spaces; mass tracking via gait analysis and biometric behavioural data; and predictions and inferences about protected attributes derived from biometric characteristics. The campaign remains active and is the international anchor for the camp-access-now-ban-biometric-surveillance-global-2021-ongoing corpus entry.

Surveillance Giants and the surveillance capitalism critique

In November 2019, Amnesty International published Surveillance Giants: How the Business Model of Google and Facebook Threatens Human Rights — the corpus's first Amnesty-authored publication entry — arguing that Google's and Facebook's surveillance-based business models are structurally incompatible with the right to privacy and pose a systemic threat to freedom of opinion and expression, freedom of thought, and equality and non-discrimination. Secretary General Kumi Naidoo framed the report's core claim at launch: "Their insidious control of our digital lives undermines the very essence of privacy and is one of the defining human rights challenges of our era." The report's strategic contribution was to translate the growing elite consensus against surveillance capitalism into a concrete three-point government-action agenda — banning surveillance advertising relying on invasive tracking and user profiling; establishing independent oversight of platform algorithms; and challenging Google's and Facebook's market dominance to enable rights-respecting alternatives. This agenda supplied a campaign and regulatory advocacy architecture that Amnesty subsequently carried into EU Digital Services Act consultation work and related regulatory debates on targeted advertising.

Stop Killer Robots and autonomous weapons

Amnesty International is a founding member and steering-committee member of the Campaign to Stop Killer Robots, the international coalition established in April 2013 to press for a new international legal instrument on autonomy in weapons systems. The coalition now comprises 250+ member organisations across 67+ countries. Amnesty's named positions in the autonomous-weapons space — that machines cannot make the complex ethical choices required for use of force; that autonomous-weapons decisions are based on biased and oppressive processes; and that fully autonomous weapons systems blur lines of accountability and undermine the right to remedy — have been the organisation's basis for supporting the Campaign's demands for a prohibition of fully autonomous weapons systems and a binding requirement for meaningful human control in the use of force.

Predictive policing and automated racism

Amnesty International UK produced the "Automated Racism" report documenting how UK police forces use data and algorithms to predict which individuals will commit crimes, finding that data-based predictive policing leads to the criminalisation, punishment, and violent policing of Black and racialised people, particularly in deprived areas. The report extends Amnesty's general algorithmic accountability framing — that AI systems reproducing structural discrimination are not neutral technical failures but a rights concern warranting prohibition or binding human-rights-compliant redesign — into the UK domestic policing context.

"Unlawful by Design" — the generative AI data pipeline challenge

In May 2026, Amnesty International published Unlawful by Design: Exposing the Human Rights Costs of Generative AI, a comprehensive analysis of the data pipelines behind six major commercial generative AI systems — GPT-3 (OpenAI), Gemini (Google), Llama (Meta), DeepSeek, Midjourney, and Stable Diffusion. The report found that the systems are built on unlawful web scraping and mass invasions of privacy by design, with data collection conducted without meaningful consent at scale; it characterised the training-data pipelines as fundamentally incompatible with international human rights law; and it called on states to prohibit standalone generative AI systems that rely on unlawful bulk data collection. The report represents Amnesty's extension of the surveillance-capitalism critique from advertising-driven platform business models (the 2019 Surveillance Giants argument) to the data-collection practices of the frontier AI industry.

AI governance summits and the military-AI accountability campaign

In February 2026, Amnesty International criticised the AI Impact Summit in New Delhi for failing to address harmful AI deployment practices by governments and technology companies. Erika Guevara Rosas, Senior Director of Research, Advocacy Policy and Campaigns, stated that "the human rights concerns arising out of technology deployment in the country were papered over." Amnesty called for binding regulations carrying "clear prohibitions around technologies that are incompatible with human rights" and "meaningful mechanisms for public participation" in shaping technological futures — continuing the organisation's pattern of critiquing voluntary-commitment AI governance frameworks as insufficient. On 5 May 2026, Amnesty International wrote to both OpenAI and Anthropic requesting information on the human rights risks of military use of generative AI; OpenAI responded on 12 May 2026, while Anthropic had not responded at the time of publication. On 15 June 2026, Amnesty International co-drafted with Access Now a joint statement on AI in Warfare, signed by civil society organisations and independent experts, calling on tech companies and states to halt the provision of AI systems for use in the military kill chain and to ensure other AI systems they provide do not cause or contribute to violations of international humanitarian law or international human rights law. The statement reflects Amnesty's view that LLM use for target generation and prioritisation pushes military operations toward forms of warfare where the foundational principles of international humanitarian law cannot be sufficiently respected.

Argentina — Sensing the Surveillance State

In August 2026, Amnesty International published Sensing the Surveillance State, documenting Argentina's expansion of AI-driven surveillance under President Javier Milei between 2024 and 2025. The report found that authorities acquired surveillance technologies valued at at least US$1.2 million — including Clearview AI facial recognition licenses, the Maltego open-source intelligence platform for collecting and organising data from social media, websites, and the dark web, and drones equipped with automated tracking, thermal cameras, and real-time image transmission — and deployed them against protesters, journalists, activists, human rights defenders, migrants, and young people. Paola Garcia Ray, Deputy Director of Amnesty International Argentina, stated that "pervasive surveillance is being used against Argentinians in both physical and digital spaces, to devastating effect." Matt Mahmoudi, Amnesty International's AI and human rights researcher, placed the findings in a wider global pattern: "What is happening in Argentina is not unique. Around the world, states engaged in authoritarian practices are ramping up their use." Amnesty called for a ban on remote biometric recognition for mass and discriminatory surveillance, regulatory frameworks governing procurement of surveillance technologies, and accountability mechanisms for abuses. The Argentina investigation is Amnesty's most detailed country-level study to date of an authoritarian government's deployment of AI surveillance as a social-control instrument, and the first to document Clearview AI's presence in a Latin American government's surveillance infrastructure.

Position in the movement graph

Amnesty International occupies the make-AI-good movement's large-international-human-rights-organisation slot — analogous to Human Rights Watch on the movement's human-rights side and to Privacy International on the data-rights side. Its Amnesty Tech programme gives the organisation a specialist identity in AI-accountability work that most organisations of comparable scale lack; the Algorithmic Accountability Lab's capacity-transfer approach and the Ban the Scan coalition-organising model are the two most concrete demonstrations of engagement with non-AI-audience communities at scale. The organisation's working theory of change — that the deployment of facial recognition, predictive policing, autonomous weapons, and surveillance-advertising systems can be constrained through a combination of grassroots-community coalition work, empirical research and public exposure, international legal frameworks, and binding regulatory demands — spans more of the movement's strategic register than most single organisations, and its resource base gives each of those tracks an institutional durability that smaller digital-rights organisations cannot maintain independently.