The European Commission, alongside national regulatory authorities, has begun with the active enforcement of key provisions under the European Union’s Artificial Intelligence (EU AI) Act.
Under these enforcement rules, interactive systems such as chatbots are now legally required to disclose to users that they are interacting with an artificial intelligence rather than a human. Additionally, all deepfakes and AI-generated or altered media (including photos, videos, and audio) must feature visible labels as well as machine-readable watermarks to prevent deception and manipulation.
For providers of general-purpose AI (GPAI) models, compliance requires documenting training procedures, maintaining a clear copyright policy, and publishing detailed summaries of training content. Providers of the most powerful models facing systemic risks must also implement measures against large-scale hazards, including cyber attacks, biosecurity risks, and loss of human control.
To support implementation, the AI Office introduced confidential Whistleblower and Complaint tools for individuals and developers, while appointing Oxford Professor Alessandro Abate as Lead Scientific Adviser alongside a 60-member independent Scientific Panel. Meanwhile, more than 180 organizations have already signed a voluntary Code of Practice to signal early compliance with these transparency standards.
Regulating Artificial Intelligence in the EU
As rapid technological advances made machine learning models increasingly capable, governments across the world recognized that traditional consumer protection and privacy frameworks were ill-equipped to handle emerging hazards such as automated discrimination, mass synthetic deception, and extreme market consolidation.
Global jurisdictions have taken noticeably different paths toward managing these risks. The European Union adopted a comprehensive, risk-based legislative framework that tiers AI applications into clear categories, banning unacceptably harmful practices like automated social scoring while enforcing rigorous audits on high-risk tools.
For more on this subject, you can see the EU AI Act coverage and on the website of the European Commission.
