Subject of the Preliminary Agreement | Key Changes
In comparison to the original Commission proposal, the key new elements of the preliminary agreement can be summarized as follows:
- Revision of the governance system with certain enforcement powers at the EU level,
- Expansion of the list of prohibitions, but with the possibility, subject to certain safeguards, of allowing law enforcement authorities to use biometric remoteidentification in public spaces,
- New protective measures to better safeguard rights: Operators of high-risk AI systems commit to conducting an impact assessment on fundamental rights before the deployment of an AI system.
Innovations in the Governance System
Due to the new regulations for GPAI models, new structures and offices are established at the EU level. These are intended to ensure effective implementation and monitoring of the new regulations for GPAI models at the EU level.
Establishment of the Office for Artificial Intelligence (AI Office)
The primary tasks of the so-called AI Office are related to the monitoring of advanced AI models, the promotion of standards and testing procedures, and the enforcement of EU regulations in all member states. The AI Office is supported by a scientific body of independent experts, which provides advisory services, particularly in the assessment and classification of basic models and the evaluation of security risks.
Artificial Intelligence Committee (AI Committee)
Furthermore, the Artificial Intelligence Committee (AI Committee) has been established. It consists of representatives from EU member states and is intended to serve as a coordination platform and advisory body for the EU Commission.
Stakeholder Consultation Forum
Stakeholders such as industry representatives, SMEs, startups, civil society, and universities/research institutions are expected to contribute technical expertise and provide advisory support to the AI Committee within the framework of the consultation forum.
Imposed Sanctions for Violations of the AI Act
The amount of fines to be paid has also changed significantly compared to the original Commission proposal. Penalties for violations of the AI Act have been set as a percentage of the global annual turnover of the infringing company in the previous fiscal year or as a predetermined amount, whichever is higher. According to this calculation, this would result in €35 million or 7% for violations related to prohibited AI applications, €15 million or 3% for violations of the obligations prescribed in the AI Act, and €7.5 million or 1.5% for providing false information. However, startups are not affected by these penalties; more proportionate limits are provided for them.
Outlook: What’s Next for the AI Act Proposal?
After the preliminary agreement, work at the technical level will continue in the coming period to finalize the details of the new regulation. Subsequently, the chair must present the compromise text for approval to the representatives of the member states (Council). The entire draft must then be confirmed by both the Council and the Parliament and undergo revision by legal and linguistic experts before being formally adopted by the two legislative bodies. The AI Act is expected to come into effect 2 years after its adoption. Given the far-reaching consequences for potential stakeholders, early clarification of all legal questions in the field of AI is advisable.
As a multidisciplinary full-service law firm, with IT law falling within our areas of expertise, we are pleased to keep you informed about the regulations regarding the use of Artificial Intelligence and their practical implementation. If you have any further questions, please do not hesitate to contact us.