Amnesty International's 2025-26 annual report was published this Tuesday morning in Dakar. This took place during a press conference organized in Dakar by Amnesty International Senegal and the Regional Office for West and Central Africa.

In the document reviewed, a section is devoted to Artificial Intelligence (AI) and human rights abuses. From the outset, Seydi Gassama and his colleagues indicated that the development of generative artificial intelligence tools, readily available to the public, continued at a breakneck pace in 2025. Furthermore, the release of numerous new models developed by tech giants like Google, Meta, and Microsoft, and major AI players such as Anthropic and OpenAI, has necessitated the construction of a vast physical infrastructure, with direct consequences for human rights.

According to the report, "it has been necessary to intensify the exploitation of natural resources, such as the minerals essential for the manufacture of computer equipment and the water used for cooling and energy consumption. "This development has had adverse consequences on human rights (environmental degradation around data centers, erosion of workers' rights)," they say.

Environmental degradation around data centers, erosion of workers' rights

 

“Faced with this situation, local struggles against the unbridled construction of data centers have multiplied in several countries, including Brazil, the United States, Ireland and Mexico. At the same time, more and more technology workers have organized to demand safer working conditions – from corporate headquarters in Silicon Valley to hubs where content moderation, data labeling and other related activities are outsourced in Africa and elsewhere,” they add.

From a regulatory standpoint, Amnesty International notes that AI "was still not sufficiently regulated" in 2025. New governance instruments adopted, such as the Indian guidelines on AI governance, are considered "non-binding or extremely general".

The organization is also concerned about the European Union's (EU) move to simplify its corporate responsibility regulation, an initiative which it believes risks weakening the European Artificial Intelligence Act, designed to ensure that AI systems are "safe, transparent, non-discriminatory and respectful of fundamental rights".

The promulgation of binding rules, applicable by States, is demanded.

 

The report focused on the issue of digital identification systems. “In 2025, the UK announced the introduction of a digital identity card, while the EU undertook to develop a digital identity wallet. According to Amnesty International, these systems have accompanied or laid the groundwork for the use of AI in the field of social protection, which has exacerbated inequalities. The integration of AI into law enforcement, immigration services, and the military has fostered human rights violations, particularly against racialized populations,” they noted.

Following these findings, Amnesty International calls on states to enact binding, enforceable, and human rights-based rules to govern AI systems. According to the organization, these rules must specifically prohibit the development and deployment of systems that are "incompatible with international human rights law."