Amnesty International's Secretary-General, Agnes Callamard, released a statement on November 27 in response to objections raised by three EU member states concerning the regulation of artificial intelligence (AI) models.
France, Germany, and Italy reached an agreement that involves not imposing overly strict regulations on the foundational models of artificial intelligence, a pivotal element of the forthcoming EU Artificial Intelligence Act.
Prior to this agreement, various technology industry players had petitioned the EU authorities, urging them to avoid excessive regulation in this nascent industry.
However, Callamard emphasized that this presents an opportunity for the region to display "global leadership" by implementing robust AI regulations. She highlighted that member states should resist succumbing to the tech industry's arguments that enacting an AI bill might lead to severe sanctions, thereby hampering innovation through stifling regulations. She challenged the notion of "innovation versus regulation," labeling it as a false dichotomy propagated by tech companies to evade meaningful accountability and binding regulatory measures.
According to Callamard, this discourse from the tech industry underscores the consolidation of power among a select group of technology companies striving to dictate the "rulebook" of artificial intelligence. Amnesty International collaborates with a coalition of civil society organizations, spearheaded by the European Digital Rights Network, advocating for EU AI laws prioritizing human rights protections.
She further highlighted that documented instances of AI's human rights violations indicate how unregulated AI systems are being exploited by states for various purposes such as welfare assessments, public space surveillance, and predicting potential criminal behavior.
Callamard urged France, Germany, and Italy to cease stalling the negotiation process, emphasizing the crucial need for EU lawmakers to concentrate on embedding significant human rights safeguards into legislation before the current EU mandate expires in 2024.
Additionally, she noted that these three countries recently joined a new set of guidelines, developed by 15 nations and major tech companies like OpenAI and Anthropic. These guidelines are aimed at providing recommendations for AI developers regarding network security practices when designing, developing, launching, and monitoring artificial intelligence models.
