Chapter 6 MEASURES IN SUPPORT OF INNOVATION Flashcards
Q: What are AI regulatory sandboxes and what is their purpose under the EU AI Act?
A: AI regulatory sandboxes are controlled environments established by Member States’ competent authorities to foster innovation and facilitate development, testing and validation of innovative AI systems before placing them on the market. They aim to improve legal certainty, support best practice sharing, foster innovation and competitiveness, contribute to regulatory learning, and facilitate market access.
Q: What are the key elements that the Commission’s implementing acts will specify regarding AI regulatory sandboxes?
A: The implementing acts will specify detailed arrangements including eligibility and selection criteria for participation, procedures for application/participation/monitoring/exit, and terms and conditions for participants. They will ensure transparency, equal access, support for SMEs, facilitation of compliance and standardization, involvement of the AI ecosystem, and streamlined processes across the EU.
Q: What are the main obligations for national competent authorities regarding AI regulatory sandboxes?
A: Authorities must establish at least one sandbox, allocate sufficient resources, provide guidance and support to identify risks, and have power to suspend testing. They must inform the AI Office and Board about establishing a sandbox, submit annual reports, and make reports publicly available. Cross-border cooperation should be facilitated.
Q: What liability provisions apply to participants in AI regulatory sandboxes?
A: Participants remain liable under applicable Union and national law for damage to third parties from sandbox experimentation. However, if providers observe the participation plan and terms and follow authority guidance in good faith, no administrative fines will be imposed for infringements of the Regulation or other supervised law.
Q: Under what conditions can personal data be processed for developing AI systems in the public interest in AI regulatory sandboxes?
A: Processing is allowed if the AI system is developed for substantial public interest in specified areas, the data is necessary to fulfil requirements, risks are monitored, data is in a separate environment, measures are taken to protect rights, and data is deleted after the sandbox. Specific conditions and national/EU laws also apply.
Q: What are the conditions for testing high-risk AI systems in real world conditions outside of AI regulatory sandboxes?
A: Testing can be conducted by providers/prospective providers before placing on market, in accordance with a testing plan approved by the market surveillance authority. Registration, oversight, data protection measures, informed consent, liability provisions and reporting obligations apply. Testing is limited to 6 months, extendable once.
Q: What is required for informed consent from subjects participating in testing high-risk AI in real world conditions?
A: Informed consent must be obtained from subjects prior to participation, after being duly informed of the nature, objectives, conditions, rights, guarantees, and arrangements of the testing. The consent must be dated and documented, with a copy provided to the subjects or their legal representatives.
Q: What measures are required to support SMEs and start-ups under the EU AI Act?
A: Member States must provide priority sandbox access for SMEs/start-ups, raise awareness, provide advice, and facilitate participation in standardization. Conformity assessment fees should be reduced proportionately for SMEs. The AI Office will provide templates, an information platform, communication campaigns, and promote best practices in AI procurement.
Q: What specific support and flexibilities are provided for micro enterprises under the EU AI Act?
A: Micro enterprises may comply with certain elements of the quality management system in Article 17 in a simplified manner, if they don’t have partner/linked enterprises. The Commission will provide guidelines on which elements can be simplified. This does not exempt micro enterprises from other requirements and obligations in the Regulation.
Q: What actions will the AI Office take to support innovation under the EU AI Act?
A: The AI Office will provide standardized templates for areas covered in the Regulation, as requested by the Board; develop a single information platform with easy-to-use information for all operators; organize communication campaigns to raise awareness of obligations; and evaluate and promote convergence of best practices in public procurement of AI systems.
