# EU AI Act The EU AI Act establishes the first comprehensive AI framework that uses a risk-based methodology to outline compliance obligations for high-risk applications and promote innovation through research exceptions and regulatory sandboxes. Status: provisional Last verified: 2026-08-27T06:04:04.000Z ## Evidence - (single_source) Article 47(1a) includes an urgency procedure exemption for law enforcement authorities without judicial authorisation - supports https://doi.org/10.55342/szemhungarica.2024.161.3.129 > . In line with the Council's mandate, the compromise text preserves the derogation from conformity assessment in Article 47(1), and also the related exemption (urgency procedure) for law enforcement authorities in Article 47(1a), without the need for judicial authorisation. However, to secure this concession, it has been necessary to re-introduce in the text the wording regarding the control of the authorisation process by the Commission, as provided in points 3, 4 and 5 of Article 47, which were previously deleted in the Council's General Approach. Concerning AI systems already placed on the market or put into service, in the case of public authorities which are providers or deployers of high-risk AI systems the compromise agreement in Article 83(2) is to offer them 4 years from the entry - (single_source) High-risk AI systems already placed on the market or put into service have a 4-year flexibility period under Article 83(2) - supports https://doi.org/10.55342/szemhungarica.2024.161.3.129 > . In line with the Council's mandate, the compromise text preserves the derogation from conformity assessment in Article 47(1), and also the related exemption (urgency procedure) for law enforcement authorities in Article 47(1a), without the need for judicial authorisation. However, to secure this concession, it has been necessary to re-introduce in the text the wording regarding the control of the authorisation process by the Commission, as provided in points 3, 4 and 5 of Article 47, which were previously deleted in the Council's General Approach. Concerning AI systems already placed on the market or put into service, in the case of public authorities which are providers or deployers of high-risk AI systems the compromise agreement in Article 83(2) is to offer them 4 years from the entry - (single_source) The EU AI Act is the first extensive legislative framework governing AI - supports https://hdl.handle.net/11511/118292 > in recruitment, performance assessment, and employee engagement. The study examines the EU AI Act, the first extensive legislative framework governing AI, and its response to these advancements via a risk-based methodology that categorizes AI systems and defines compliance requirements for high-risk applications, particularly in employment settings. The study integrates a legal examination of the EU AI Act with an assessment of practical applications, business methodologies, and technology-oriented trends in human resources. It utilizes official legislation, institutional reports, professional analyses, and AI tool outputs to highlight both the possible advantages-such as enhanced efficiency and strategic decision-making-and the hazards, including bias, opacity, and data misuse. The study - (single_source) The Council's mandate preserves Article 47(1) derogation from conformity assessment - supports https://doi.org/10.55342/szemhungarica.2024.161.3.129 > . In line with the Council's mandate, the compromise text preserves the derogation from conformity assessment in Article 47(1), and also the related exemption (urgency procedure) for law enforcement authorities in Article 47(1a), without the need for judicial authorisation. However, to secure this concession, it has been necessary to re-introduce in the text the wording regarding the control of the authorisation process by the Commission, as provided in points 3, 4 and 5 of Article 47, which were previously deleted in the Council's General Approach. Concerning AI systems already placed on the market or put into service, in the case of public authorities which are providers or deployers of high-risk AI systems the compromise agreement in Article 83(2) is to offer them 4 years from the entry - (single_source) The compromise text re-introduces Commission control of the authorisation process by referencing points 3, 4 and 5 of Article 47 - supports https://doi.org/10.55342/szemhungarica.2024.161.3.129 > . In line with the Council's mandate, the compromise text preserves the derogation from conformity assessment in Article 47(1), and also the related exemption (urgency procedure) for law enforcement authorities in Article 47(1a), without the need for judicial authorisation. However, to secure this concession, it has been necessary to re-introduce in the text the wording regarding the control of the authorisation process by the Commission, as provided in points 3, 4 and 5 of Article 47, which were previously deleted in the Council's General Approach. Concerning AI systems already placed on the market or put into service, in the case of public authorities which are providers or deployers of high-risk AI systems the compromise agreement in Article 83(2) is to offer them 4 years from the entry - (single_source) The EU AI Act outlines compliance requirements for high-risk applications - supports https://hdl.handle.net/11511/118292 > in recruitment, performance assessment, and employee engagement. The study examines the EU AI Act, the first extensive legislative framework governing AI, and its response to these advancements via a risk-based methodology that categorizes AI systems and defines compliance requirements for high-risk applications, particularly in employment settings. The study integrates a legal examination of the EU AI Act with an assessment of practical applications, business methodologies, and technology-oriented trends in human resources. It utilizes official legislation, institutional reports, professional analyses, and AI tool outputs to highlight both the possible advantages-such as enhanced efficiency and strategic decision-making-and the hazards, including bias, opacity, and data misuse. The study - (single_source) The proposed AI Act ensures privacy through its risk-based approach - supports http://lup.lub.lu.se/student-papers/record/9121353 > Artificial intelligence (AI) within healthcare creates opportunities to save more human lives regarding the prevention and prediction of diseases. This thesis has its main focus on the proposed AI Act and in what way the proposed AI Act promotes innovation and ensures privacy concerning the collection of patient data for the prevention and prediction of diseases. ... This thesis concludes that the proposed AI Act does promote innovation through research exceptions, regulatory sandboxes, and removing barriers for Small and Medium Enterprises (SMEs). The proposed AI Act also ensures privacy through its risk-based approach. ... . One argument is that the proposed AI Act does hinder the innovation of AI because the focus of the risk-based approach is too highly valued. Another argument is that - (single_source) The proposed AI Act promotes innovation through research exceptions, regulatory sandboxes, and removing barriers for Small and Medium Enterprises - supports http://lup.lub.lu.se/student-papers/record/9121353 > Artificial intelligence (AI) within healthcare creates opportunities to save more human lives regarding the prevention and prediction of diseases. This thesis has its main focus on the proposed AI Act and in what way the proposed AI Act promotes innovation and ensures privacy concerning the collection of patient data for the prevention and prediction of diseases. ... This thesis concludes that the proposed AI Act does promote innovation through research exceptions, regulatory sandboxes, and removing barriers for Small and Medium Enterprises (SMEs). The proposed AI Act also ensures privacy through its risk-based approach. ... . One argument is that the proposed AI Act does hinder the innovation of AI because the focus of the risk-based approach is too highly valued. Another argument is that - (single_source) The EU AI Act categorizes AI systems via a risk-based methodology - supports https://hdl.handle.net/11511/118292 > in recruitment, performance assessment, and employee engagement. The study examines the EU AI Act, the first extensive legislative framework governing AI, and its response to these advancements via a risk-based methodology that categorizes AI systems and defines compliance requirements for high-risk applications, particularly in employment settings. The study integrates a legal examination of the EU AI Act with an assessment of practical applications, business methodologies, and technology-oriented trends in human resources. It utilizes official legislation, institutional reports, professional analyses, and AI tool outputs to highlight both the possible advantages-such as enhanced efficiency and strategic decision-making-and the hazards, including bias, opacity, and data misuse. The study ## Timeline - 2026-08-27T06:00:27.000Z: The EU AI Act establishes the first comprehensive AI framework The legislation employs a risk-based methodology to categorize AI systems and mandates compliance requirements for high-risk applications Article 47(1a) includes an urgency procedure exemption for law enforcement authorities without judicial authorisation The compromise text re-introduces Commission control of the authorisation process by referencing points 3, 4 and 5 of Article 47 Article 47(1) derogation preserves conformity assessment for the Council's mandate High-risk AI systems placed on the market or into service have a 4-year flexibility period under Article 83(2) The proposed AI Act promotes innovation through research exceptions, regulatory sandboxes and SME-friendly measures The proposed AI Act ensures privacy via a risk-based approach