| Abstract [eng] |
The thesis examines an inconsistency between the European Union’s (EU) framing of artificial intelligence (AI) as primarily a technical and commercial issue, alongside its emphasis on risks. It analyses how the EU constructs AI-related security knowledge in its emerging AI policy, drawing on discourse analysis. Using the concept of ‘riskification’ – including the referent object, conditions of possibility for harm, governance programme, and international engagement – it finds that fundamental rights and democratic governance emerge as referent objects. Potential harm is framed as intrusion, discrimination, and autonomy representing AI as Other. The EU responds with diverse policy measures and internationally seeks influence through multilateral engagement while competing with international actors. The thesis introduces three conceptualizations: agentic security, technocratization, and a fortress. These reflect a focus on protecting human agency and the capacity to sustain control, decision making, and the hierarchy of the superior human position, while a technocratized governance programme transforms security into routine practices. The EU’s positionality as a fortress aims for enhanced self-protection and decreased interdependence with international powers. Together, these concepts reveal controversies of anthropocentrism, depoliticization and Eurocentrism, raising questions about their correspondence to the evolving challenges of human-machine interactions. |
