EU AI Labels Arrive, but Enforcement Still Has Loopholes
EU AI Labels Arrive, but Enforcement Still Has Loopholes
Europe’s attempt to make AI easier to spot has entered a tougher phase: new disclosure duties are now live, but the practical timetable is less uniform than the headline suggests.
From August 2, providers must tell users when they are dealing with an AI system, unless that is already obvious, and attach machine-readable markers to synthetic audio, images, video and text. Platforms deploying realistic AI-manipulated media must label deepfakes. The European Commission’s rationale is straightforward: people should be able to judge what they are seeing, calibrate their trust and avoid deception.
The rules are being framed as AI’s “cookie banner” moment—a push to turn a complicated technical obligation into a visible, routine online disclosure. The Commission has offered optional common icons in an effort to avoid a proliferation of incompatible platform labels. But the underlying labeling obligation is mandatory, and failures can carry penalties of up to €15 million or 3% of global annual turnover.
Meta illustrates the tension. The company has signed a voluntary EU code on transparency while warning that a growing array of labels and disclosures could confuse users and regulators—despite having introduced its own bespoke labeling system. Its position reflects an industry concern about fragmentation; regulators, by contrast, are seeking a baseline that works across services.
Enforcement will also be divided. The AI Office oversees general-purpose model providers and systems offered by the same provider; national authorities handle other systems, while the European Data Protection Supervisor covers EU institutions. Complaint and whistleblower channels are now available. For older systems, however, compliance is not entirely immediate: the Digital Omnibus extended the watermarking grace period to December 2, 2026, while some high-risk rules were pushed back further.
Continue reading https://foxvector.com/stories/019fca8b-46b3-02cf-73c3-03f2afbfb5f2
Write a comment