The EU AI Act must address human rights concerns, urged human rights organisations
The proposed EU AI Act has been criticised by human rights organisations for failing to ban many harmful and dangerous uses of AI in the context of immigration enforcement. Data-intensive technologies, including AI systems, are increasingly being used to make Europe’s borders impenetrable.
The European Parliament plans to propose stricter rules for foundation models, such as ChatGPT, under the AI Act to regulate AI based on its capacity to cause harm. The proposed rules include compliance requirements for foundation model providers, data governance measures, and transparency obligations. Downstream economic operators would become responsible for complying with the AI Act’s stricter regime if they modify a high-risk model.
However, the proposed EU AI Act has been criticised by human rights organisations for failing to ban many harmful and dangerous uses of AI in the context of immigration enforcement. Data-intensive technologies, including AI systems, are increasingly being used to make Europe’s borders impenetrable, which pushes people towards more precarious and deadly routes, strips them of their fundamental privacy rights and unjustifiably prejudices their claims to immigration status. The human rights organisations name the European border agency Frontex as an example, which stands accused of being complicit in grave human rights violations at many EU borders and is known to use various AI-powered technological systems to facilitate illegal pushback operations. Therefore, human rights organisations have called on the EU lawmakers to ensure that the legislation protects everyone, including asylum seekers, from dangerous and racist surveillance technologies and to ensure that AI technologies are used to protect, not surveil.
The last few days have seen increased calls for stricter regulation of AI. A group of 12 European Union lawmakers working on legislation related to AI have called for a summit to discuss ways to control the development of advanced AI systems, stating that they were evolving faster than expected. In addition, forty-two German trade unions and associations have urged the European Union to strengthen draft AI rules due to concerns about generative AI, such as ChatGPT. ARTICLE 19, a human rights organisation, called for a ban on remote biometric surveillance and emotion recognition technologies in the AI Act.
The European Commission will finalise the details of AI rules over the coming months before they become legislation. The political agreement on the AI Act will be voted on by leading European Parliament committees on 26 April.