site.btaNew EU AI Transparency Rules to Enter into Force on August 2

New European Union transparency obligations for AI providers and deployers will enter into force on August 2, the European Commission press service announced on Friday.

Providers of certain AI systems, including generative and interactive AI systems, as well as providers of deepfake content, will be obliged to inform individuals whenever they are directly interacting with an AI system. Starting August 2, deepfakes (AI-generated or manipulated images, audio, or video content) will be subject to labelling rules. Providers must apply a machine-readable mark to synthetic content generated or manipulated by AI and enable its detection.

The EU regulations seek to support the safe implementation, use and further development of AI, so that the technology enjoys greater public trust, the press release noted.

Providers of General-Purpose AI (GPAI) models will have to fulfill additional obligations to manage the risks of damage caused by chemical, biological, radiological, and nuclear incidents, loss of control, cybercrime, harmful manipulation, and threats to fundamental rights. They will be required to assess threats to European cybersecurity, as well as risks associated with AI systems operating independently of humans. The providers will be expected to implement a copyright policy and to publish a comprehensive summary of the content used for GPAI model training.

The European Union introduced transparency obligations related to prohibited AI practices, including systems that exploit individuals' vulnerabilities to cause harm.

Three authorities will be responsible for the regulations' enforcement: the AI Office (for systems under its supervision, including those built into EU-defined very large online platforms or very large online search engines), the European Data Protection Supervisor (when EU institutions are providers/deployers) and the national market surveillance authorities (for other AI systems).

Individuals and legal entities will have the right to file complaints if they suspect violations of the EU AI Act. GPAI providers’ employees will be able to report irregularities and address alleged infringements, the European Commission stressed.