EU Mandates AI Content Labels. Deepfakes Are Now Better Labeled and Harder to Detect.

The EU now requires labels on AI-generated content. A deepfake is labeled correctly as potentially AI-generated. The deepfake is more convincing than before. The label is applied. The label is compliant. The deepfake is still a deepfake.

Regulatory capture often works by splitting governance. One office handles labeling. Another handles detection. They operate under different metrics. Labeling is about disclosure. Detection is about prevention. A label proves nothing about what lies beneath it. The two offices know this about each other. They have not synchronized.

The deepfake circulates. It is labeled. It is believed. The label satisfied the regulation. The regulation satisfied the government. The deepfake satisfied itself by existing. Everyone was adequate.