Artificial intelligence fuels global tensions, threatens fundamental rights, warns Amnesty
Amnesty International warns that artificial intelligence, often promoted as a solution for security and anti-discrimination efforts, actually exacerbates racism and other forms of discrimination, thus undermining fundamental human rights.
“Artificial intelligences are touted as effective solutions for necessary security measures and combating discrimination. However, instead of solving these issues, they amplify racism and discrimination and perpetuate breaches of fundamental rights,” said Katia Roux, advocacy officer for technology and human rights at Amnesty International, during the presentation of the annual “State of the World’s Human Rights” report April 24.
These dangers are heightened by two main factors: a critical election year, with over half of the world’s population called to vote, and an international balance disrupted by numerous high-intensity conflicts in places like Gaza, Ukraine, and Sudan. This precarious situation benefits AIs, which thrive in the absence of stringent regulation and feed these tensions.
Discriminations fueled by algorithms
According to Amnesty International, different AIs can have varied impacts depending on their use but contribute to racial or ethnic discrimination. “For instance, during the conflict in Ethiopia between 2020 and 2022, as well as in Myanmar and with the Rohingya, it was observed that Meta’s (Facebook) algorithms amplified violent content that incited hatred against the Tigrayans. This emphasis on engaging violent content leads to surges of hate,” analyzed Roux.
Beyond the algorithms of digital giants, newer technologies such as generative AIs – which can independently create content from a simple instruction – also disrupt political life. They do not spare democracies either. For example, in the United Kingdom, which is expected to have elections in January 2025, a series of “deepfakes” (images generated by artificial intelligence) of Labour MP and opposition leader Keir Starmer insulting employees were distributed, disrupting politics in a new way.
A similar incident occurred in the United States during the New Hampshire Democratic primary in January. A few months before the November presidential election, the campaign manager for Joe Biden’s rival admitted to using AI to create a phone call in which the American president urges his voters not to vote against him. This fabricated audio disrupted the early primaries.
According to this latest Amnesty International report, facial recognition and emotion detection tools, which are also forms of artificial intelligence, further fuel unequal treatment, even if used by governments.
Europe, for instance, was criticized by Amnesty International for using the “iBorder Control” software, which acts like a lie detector during border checks. Used at Latvian, Hungarian, and Greek borders and funded by the European Union, this system is said to favor discrimination, the NGO argues.
“Borders are interesting for governments because they can test their tools on vulnerable people,” said Roux. “iBorder Control, like others, is labeled as ‘neutral’ technology – a mere aid to human decision-making – but they are not because they are designed by humans who can choose the criteria and can be abusively used to target specific groups.”
Missed opportunity with the AI Act
The widespread use of artificial intelligence for security issues, whose effectiveness and necessity have not been demonstrated, according to Amnesty International, is subject to very little regulation. The European Parliament, however, adopted March 24 the Artificial Intelligence Act, a regulation establishing rules on artificial intelligence across all areas, whether algorithmic, security, or creative. According to Roux, this regulation “does not go far enough. It was the first time in the world that we had a regulation addressing all these points, and it is a pity that some states did not pursue an ambitious project, especially regarding facial identification.”
As the Olympic Games approach, France has become the first country in the European Union to legalize the use of artificial intelligence for surveillance purposes. This move concerns Amnesty International, even though the Ministry of the Interior assures that its use should be limited in time. Amnesty International fears that these systems may become a permanent fixture in everyday life.
