WHAT TO FIX’s Submission to Article 50 AI Act Consultation

In our latest submission to the European Commission, we argue for AI labeling obligations, arguing that they can create signals that can be used to detect manipulative monetization strategies and inauthentic behavior at scale.

This becomes especially relevant when AI-generated content intersects with recommender systems and monetization incentives. Social media business models often reward high-engagement content regardless of authenticity. AI systems substantially reduce the cost of producing emotionally manipulative material at scale, enabling actors to optimize outputs for virality, outrage, or persuasion. Transparency obligations therefore create the possibility of linking inauthentic content (synthetic media) with inauthentic behavior (coordinated amplification, deceptive engagement practices, monetized spam networks, or disinformation campaigns).

The combination of machine-readable labeling under Article 50(2) and deployer disclosure duties under Article 50(4) could enable platforms and regulators to develop risk indicators around monetization services financially incentivizing inauthentic behavior and content. For instance, repeated circulation of labelled AI-generated political content through monetized accounts and content could support detection of disinformation infrastructures, particularly where synthetic media is combined with coordinated amplification tactics.