Morocco's Call for a New Accountability Framework in AI
On July 7, 2026, during a high-level session in Geneva, Aml Falah Sagroshni, the Minister Delegate to the Head of Government in charge of Digital Transition and Administrative Reform, emphasized the urgent need for establishing a new accountability model for AI systems. She highlighted that the responsibility of these systems represents "perhaps one of the greatest legal challenges of this decade." The minister's remarks were part of a broader discussion on the importance of upholding and protecting human rights in the age of artificial intelligence, held within the framework of the global dialogue on AI governance.
During her presentation, Sagroshni explained that AI systems often function through a network of independent agents, each contributing to the decision-making process without a single human entity directly accountable for the final outcome. This decentralized approach complicates the identification of responsibility when malfunctions occur, as it can be challenging to ascertain which agent should be held accountable or how to distribute liability among the various involved parties.
She further elaborated that the collaborative nature of these systems means that even if multiple actors contribute to a process, the end result could be detrimental without any clear indication of where the error occurred or the timing of the mistake. This complexity is exacerbated by issues surrounding temporal causality and the intricate models composed of billions of variables. To illustrate this dilemma, Sagroshni pointed out that Morocco handles approximately 52 million administrative transactions annually, stressing that accurately pinpointing the source of an error within such a vast operation poses a significant challenge.
Developing New Approaches to AI Accountability
In light of these challenges, the minister argued for the necessity of developing new methodologies that can effectively address the growing complexity of AI systems. She advocated for an approach based on "decoupling," which would allow for the reconstruction of a non-deterministic causal diagram of the actions undertaken by agent systems. Furthermore, Sagroshni outlined three essential commitments to enhance the accountability of AI systems: systematically documenting the chains of agents to clarify who does what, how, and when—particularly within public services; designating a clearly defined human responsibility for each decision impacting individuals' rights; and ensuring the right to effective remedy within a timeframe that aligns with the operational speed of AI systems.
Sagroshni stressed that "the operational independence of algorithms must never translate into independence from the law." She also called for the adoption of proactive governance grounded in human rights, which is adaptable to the increasing complexity of AI technologies. This includes the development of practical mechanisms for risk assessment, traceability, auditing, and human oversight, while considering the institutional specifics and developmental priorities of each country. The discussions during this session emphasized key issues such as transparency, accountability, human oversight, and risk prevention throughout the lifecycle of AI systems, as well as the protection of vulnerable groups, including children and women.
As reported by maroc.ma.
