Med Kharbach and Johanathan Woodworth of Mount Saint Vincent University examine the EU AI Act’s Article 50 requirement, taking full effect by December 2026, that generative AI systems embed machine-readable watermarks in their outputs. They argue invisible watermarks in personal messages could damage interpersonal trust because they reveal AI use only after the fact and cannot distinguish minor AI-assisted editing from full AI composition. The authors describe this as a ‘verdict delivered after the fact’ that breeds suspicion in everyday communication, and argue voluntary, contextual disclosure between people is preferable to after-the-fact technical detection.
