The European Union announces the expansion of the implementation of the Artificial Intelligence Act

Brussels, July 31 (Xinhua) -- The European Commission announced on the 31st that the EU will expand the scope of application of the Artificial Intelligence Law from August 2, and begin to implement transparency rules and relevant rules for general artificial intelligence model providers.

The European Commission released a statement on the 31st stating that under new transparency rules, interactive artificial intelligence systems such as chatbots must inform users that they are interacting with artificial intelligence rather than humans. Images, videos, audio, and other "deepfake" content generated or modified using artificial intelligence must be identified. The content generated or modified by artificial intelligence must also have machine-readable tags for easy recognition. The European Commission stated that these regulations aim to reduce deception and manipulation.

At the same time, starting from August 2nd, the European Artificial Intelligence Office can enforce relevant rules of the Artificial Intelligence Act on general artificial intelligence model providers. For example, for advanced general artificial intelligence models that may pose systemic risks, their providers must also fulfill additional obligations to prevent the risk of large-scale harm, such as cyber attacks, model loss of control, harmful manipulation, and infringement of basic rights.

The EU's Artificial Intelligence Law came into effect on August 1, 2024, categorizing artificial intelligence systems into unacceptable risk, high risk, transparency risk, and minimum or no risk levels based on the potential risks they may pose. The law adopts a phased application arrangement, with most provisions applicable from August 2, 2026.

In addition, the EU has postponed the application of high-risk AI system rules for recruitment, education, law enforcement, and immigration management through the "Comprehensive Simplification Package" regarding artificial intelligence until December 2, 2027; The rules for high-risk artificial intelligence systems embedded in regulated products such as medical devices and machinery will be postponed until August 2, 2028.