Subscribe
Starting August 2, 2026, the European Union will enter a new phase in the regulation of artificial intelligence with the launch of the operational application of the AI Act. Oversight will be entrusted to the European Commission’s AI Office, which will work in collaboration with the competent national authorities to ensure compliance with the provisions of the regulation. At the same time, new transparency requirements will come into force, designed to strengthen citizens’ and businesses’ trust in AI-based technologies and to ensure the safe development and use of such systems throughout the Union.
TRANSPARENCY REQUIREMENTS FOR ARTIFICIAL INTELLIGENCE SYSTEMS
Chatbots and other conversational systems will have to clearly inform users that the interlocutor is an automated system and not a natural person. A similar principle applies to multimedia content produced or modified using artificial intelligence techniques. Images, videos, and audio files classified as deepfakes must be clearly identified as such. Furthermore, all content generated or altered by artificial intelligence must be accompanied by machine-readable markers, facilitating identification and recognition. The objective of these requirements is to limit deception, manipulation, and misinformation, enabling users to make informed choices. At the same time, the new provisions provide companies with clearer operational guidance regarding the requirements and the tools they can use to demonstrate compliance with the regulation.
MANAGING SYSTEMIC RISK AND PROHIBITING THE MOST HARMFUL PRACTICES
The regulation introduces additional obligations for the most advanced artificial intelligence models, considered potentially capable of generating systemic risks. Their providers will be required to adopt specific measures to prevent large-scale damage resulting from the use of these technologies. The risks addressed include those related to chemical, biological, radiological, and nuclear accidents, loss of control over artificial intelligence systems, cyberattacks, malicious manipulation, and threats to fundamental rights. The legislation also addresses issues that have recently emerged in public debate, such as risks to European cybersecurity and the possibility of artificial intelligence systems operating outside of human control. At the same time, enforcement begins of rules prohibiting certain artificial intelligence practices deemed incompatible with the fundamental values of the European Union. These include systems capable of manipulating people’s behavior, maliciously exploiting their vulnerabilities, or making discriminatory or unjustified assessments that could compromise the exercise of fundamental rights.
INSTITUTIONAL SUPERVISION STRUCTURE AND SCIENTIFIC SUPPORT
The implementation of the new rules is based on a division of responsibilities among various bodies. The AI Office supervises systems developed by the same provider as the underlying AI model and systems integrated into very large online platforms or large search engines designated under the Digital Services Act. The competent national authorities are responsible for overseeing other AI systems, while the European Data Protection Supervisor supervises the applications used by European Union institutions, bodies, and agencies. The overall effectiveness of the system will also depend on the ability of Member States to promptly designate competent authorities and ensure they have adequate organizational and financial resources. The supervisory activity will be supported by a Scientific Panel composed of sixty independent experts in the field of artificial intelligence, an advisory body that recently held its first meeting. This structure is complemented by the appointment of Professor Alessandro Abate, a lecturer in the Department of Computer Science at the University of Oxford, as the AI Office’s Lead Scientific Adviser. His role involves supporting scientific activities related to general-purpose artificial intelligence models, with a particular focus on innovation, dissemination, testing, and performance evaluation processes.
REPORTING TOOLS AND TIMETABLE FOR FUTURE PROVISIONS
To facilitate enforcement of the legislation, the AI Office has developed several tools for citizens, businesses, and industry professionals. Individuals and legal entities can use a dedicated system to submit reports of alleged violations of the AI Act by suppliers under the Office’s jurisdiction. Workers employed by suppliers of artificial intelligence systems or models also have a dedicated whistleblowing channel through which they can safely report potential violations. Finally, an additional channel has been established for suppliers developing systems based on general-purpose artificial intelligence models who wish to report any violations committed by the suppliers of the models they use. All information transmitted through these tools will be treated confidentially. The AI Act’s implementation timeline has since been amended by the AI Omnibus. The provisions regarding high-risk AI systems will enter into force on December 2, 2027, while those regarding high-risk systems integrated into regulated products will apply from August 2, 2028. The same regulatory intervention introduces further prohibitions regarding AI systems intended for the production of sexually explicit content without consent and material involving child sexual abuse. These new prohibitions will become effective starting December 2, 2026, gradually completing the European regulatory framework aimed at ensuring the use of AI in compliance with the principles of safety, transparency, and protection of fundamental rights.