AI Risks and Sovereignty: What Mexico's Institutional Agenda Recognises and Addresses

Countries that adopt advanced artificial intelligence systems without controlling their development face asymmetries between the jurisdictions where technological decisions are made and those where their consequences are felt. This study examines which risks Mexico's institutional agenda recognises and how it responds to AI, considering the relationship of these actions to direct and indirect sovereignty and the extent to which they have been carried out or are in force. To this end, it develops an AI Risk Map for Mexico, comprising seven scenarios and their respective risk pathways, and constructs an original dataset of 192 legislative, judicial and institutional actions drawn from Observatorio IA México and cross-checked against official sources. The findings show an agenda concentrated mainly on malicious AI use (23.4%) and the use of public and private data (22.9%), while measures concerning oversight of systems and providers show limited levels of realisation. Of the 80 actions directly or indirectly related to sovereignty, only ten (12.5%) have been carried out or are in force. The analysis identifies a distinction between risks recognised, institutional responses formulated and measures actually carried out, without equating their legal force with effectiveness. It also proposes a framework that can be applied to other countries to compare institutional agendas and identify shared vulnerabilities and opportunities for international coordination.