The AI Act

The AI Act also recognises the concept of general purpose AI (GPAI); AI systems that can be used for multiple purposes and are usually generative in nature, ie creating output content based on user inputs or 'prompts'. Specific rules apply to GPAI models, particularly those with high-impact capabilities. Such models are subject to stricter requirements due to their potential systemic risks and significant market impacts.

The AI Act provides a legislative framework to regulate entry into the EU internal market and mandates conformity assessments to verify AI systems' compliance with established standards. These assessments can be carried out in two ways: either self assessments by AI system providers, or third-party assessments conducted by designated notifying bodies. Notifying bodies also have auditing authority to ensure proper conformity.

The AI Act creates a number of new advisory and regulatory bodies including:

• The AI Office: attached to the European Commission, the AI Office co-ordinates the implementation of the AI Act across Member States and oversees the compliance of general purpose AI providers.
• The European Artificial Intelligence Board: comprising representatives from each Member State, the Board advises the Commission and facilitates the application of the AI Act, fulfilling a similar role to that of the European Data Protection Board under the GDPR.
• The Advisory Forum: this forum, consisting of a balanced selection of stakeholders from industry, start-ups, SMEs, civil society, and academia, provides technical expertise and advice to the Board and the Commission.
• National competent authorities: Member States must designate national competent authorities responsible for implementing the AI Act and conducting market surveillance. They verify AI systems' compliance, oversee conformity assessments, and appoint third parties for external conformity assessments.