1
00:00:00,050 --> 00:00:04,070
Lesson, notification and enforcement mechanisms under the UI act.

2
00:00:04,100 --> 00:00:10,940
Notification and enforcement mechanisms under the UI act are critical components in ensuring compliance

3
00:00:10,940 --> 00:00:15,590
and fostering trust in artificial intelligence systems across the European Union.

4
00:00:15,860 --> 00:00:22,160
The UI act, formerly known as the Artificial Intelligence Act, aims to create a robust regulatory

5
00:00:22,160 --> 00:00:28,790
framework to address the risks associated with AI and to promote the uptake of trustworthy AI while

6
00:00:28,790 --> 00:00:31,460
ensuring the protection of fundamental rights.

7
00:00:32,660 --> 00:00:38,930
Notification mechanisms under the EU AI act primarily involve the requirement for AI system providers

8
00:00:38,930 --> 00:00:44,630
to inform relevant authorities and stakeholders about the deployment and use of their AI systems.

9
00:00:45,230 --> 00:00:50,600
These mechanisms are designed to provide transparency and accountability, ensuring that AI systems

10
00:00:50,600 --> 00:00:52,850
are used responsibly and ethically.

11
00:00:53,750 --> 00:01:00,020
Providers of high risk AI systems must register their systems with a designated database managed by

12
00:01:00,020 --> 00:01:01,310
the European Commission.

13
00:01:01,490 --> 00:01:08,060
This registration process includes detailed information about the AI system, such as its intended purpose,

14
00:01:08,090 --> 00:01:12,320
the data used for training, and the measures taken to mitigate risks.

15
00:01:13,580 --> 00:01:19,700
The notification process also requires providers to conduct a conformity assessment before placing high

16
00:01:19,700 --> 00:01:21,770
risk AI systems on the market.

17
00:01:22,610 --> 00:01:28,220
This assessment involves evaluating the AI systems compliance with the requirements set out in the EU

18
00:01:28,250 --> 00:01:33,470
AI act, including risk management, data governance, and human oversight.

19
00:01:33,830 --> 00:01:39,080
Providers must document the results of the conformity assessment and make them available to the relevant

20
00:01:39,080 --> 00:01:40,850
authorities upon request.

21
00:01:41,240 --> 00:01:47,660
This documentation ensures that AI systems are thoroughly evaluated for safety and ethical considerations

22
00:01:47,660 --> 00:01:49,160
before they are deployed.

23
00:01:50,600 --> 00:01:57,350
Enforcement mechanisms under the EU AI act are essential for ensuring compliance and addressing violations.

24
00:01:57,710 --> 00:02:03,340
The act establishes a framework for monitoring and enforcement at both the national and EU levels.

25
00:02:03,430 --> 00:02:09,310
National competent authorities are responsible for supervising the implementation of the AI act within

26
00:02:09,310 --> 00:02:11,200
their respective jurisdictions.

27
00:02:11,740 --> 00:02:17,350
These authorities have the power to conduct inspections, request information, and impose penalties

28
00:02:17,350 --> 00:02:18,880
for non-compliance.

29
00:02:20,560 --> 00:02:26,800
At the EU level, the European Artificial Intelligence Board plays a crucial role in coordinating enforcement

30
00:02:26,800 --> 00:02:28,420
efforts across member states.

31
00:02:29,260 --> 00:02:35,530
The EIB is tasked with facilitating cooperation between national authorities, providing guidance on

32
00:02:35,530 --> 00:02:41,260
the interpretation and application of the AI act, and promoting the exchange of best practices.

33
00:02:42,070 --> 00:02:47,980
The board also has the authority to issue recommendations and opinions on specific cases, ensuring

34
00:02:47,980 --> 00:02:51,040
a harmonised approach to enforcement across the EU.

35
00:02:52,120 --> 00:02:57,850
Penalties for non-compliance with the EU I act can be substantial, reflecting the seriousness of the

36
00:02:57,850 --> 00:03:00,580
violations and the potential harm caused.

37
00:03:01,000 --> 00:03:07,360
For example, fines of up to 6% of the total worldwide annual turnover of the preceding financial year,

38
00:03:07,360 --> 00:03:13,300
or €30 million, whichever is higher, can be imposed for breaches of the Act's provisions related to

39
00:03:13,330 --> 00:03:17,980
prohibited I practices and obligations for high risk AI systems.

40
00:03:18,550 --> 00:03:23,500
These stringent penalties underscore the importance of adhering to the regulatory requirements, and

41
00:03:23,500 --> 00:03:28,840
demonstrate the EU's commitment to ensuring the safe and ethical use of AI.

42
00:03:30,160 --> 00:03:35,470
One notable aspect of the enforcement mechanism is the role of market surveillance authorities.

43
00:03:35,920 --> 00:03:41,440
MSAs are responsible for monitoring AI systems on the market and ensuring their compliance with the

44
00:03:41,440 --> 00:03:42,310
AI act.

45
00:03:42,640 --> 00:03:48,100
They have the authority to carry out on site inspections, request access to technical documentation,

46
00:03:48,100 --> 00:03:50,440
and perform testing of AI systems.

47
00:03:51,040 --> 00:03:57,100
If an MSA identifies a non-compliant AI system, it can take corrective actions such as ordering the

48
00:03:57,100 --> 00:04:00,160
withdrawal or recall of the system from the market.

49
00:04:00,640 --> 00:04:06,060
This proactive approach to market surveillance helps to prevent the dissemination of harmful or unethical

50
00:04:06,090 --> 00:04:09,180
AI systems and protects consumers and users.

51
00:04:11,190 --> 00:04:16,830
The EU AI act also emphasizes the importance of international cooperation in enforcement.

52
00:04:17,100 --> 00:04:22,740
The act recognizes that AI systems often operate across borders, and that effective enforcement requires

53
00:04:22,740 --> 00:04:25,710
collaboration with authorities outside the EU.

54
00:04:26,070 --> 00:04:31,200
To this end, the European Commission is encouraged to engage in dialogue and cooperation with international

55
00:04:31,200 --> 00:04:37,500
organizations and third countries to promote the alignment of AI regulations and facilitate the enforcement

56
00:04:37,500 --> 00:04:39,360
of the AI act globally.

57
00:04:42,090 --> 00:04:48,660
In addition to formal enforcement mechanisms, the EU AI act promotes the use of regulatory sandboxes

58
00:04:48,660 --> 00:04:51,720
to foster innovation while ensuring compliance.

59
00:04:51,810 --> 00:04:57,270
Regulatory sandboxes are controlled environments where AI developers can test their systems under the

60
00:04:57,270 --> 00:04:59,670
supervision of regulatory authorities.

61
00:04:59,940 --> 00:05:06,810
These sandboxes provide a safe space for experimentation and allow developers to identify and address

62
00:05:06,810 --> 00:05:11,070
potential compliance issues before bringing their AI systems to market.

63
00:05:11,070 --> 00:05:17,310
By supporting innovation within a regulated framework, regulatory sandboxes help to balance the need

64
00:05:17,310 --> 00:05:22,740
for technological advancement with the imperative of safety and ethical considerations.

65
00:05:23,700 --> 00:05:29,610
The notification and enforcement mechanisms under the EU AI act are designed to create a comprehensive

66
00:05:29,610 --> 00:05:36,240
regulatory framework that addresses the risks associated with AI, while promoting its responsible and

67
00:05:36,240 --> 00:05:37,290
ethical use.

68
00:05:37,830 --> 00:05:42,120
By requiring transparency and accountability through notification processes.

69
00:05:42,120 --> 00:05:48,630
The act ensures that AI system providers take their responsibilities seriously and adhere to the highest

70
00:05:48,630 --> 00:05:50,700
standards of safety and ethics.

71
00:05:51,030 --> 00:05:57,840
The robust enforcement mechanisms, including the roles of national authorities, the EIB and MSAs,

72
00:05:57,840 --> 00:06:03,540
provide the necessary oversight and ensure that violations are promptly addressed and penalized.

73
00:06:03,990 --> 00:06:09,650
Furthermore, the emphasis on international cooperation and the use of regulatory sandboxes demonstrates

74
00:06:09,650 --> 00:06:13,970
the EU's commitment to fostering a global ecosystem of trustworthy AI.

75
00:06:14,780 --> 00:06:20,810
By aligning AI regulations and facilitating cross-border enforcement, the EU aims to create a level

76
00:06:20,810 --> 00:06:26,840
playing field for AI innovation while protecting fundamental rights and promoting public trust in AI

77
00:06:26,870 --> 00:06:27,980
technologies.

78
00:06:29,360 --> 00:06:35,840
In conclusion, the notification and enforcement mechanisms under the UI act are integral to the Act's

79
00:06:35,870 --> 00:06:41,570
overarching goal of ensuring the safe, ethical and responsible use of AI in the EU.

80
00:06:42,410 --> 00:06:47,990
These mechanisms provide a comprehensive framework for transparency, accountability, and oversight,

81
00:06:47,990 --> 00:06:53,720
fostering an environment where AI can thrive while safeguarding the rights and interests of individuals

82
00:06:53,720 --> 00:06:55,340
and society as a whole.

83
00:06:55,340 --> 00:07:01,460
The EU AI act sets a high standard for AI governance and serves as a model for other jurisdictions seeking

84
00:07:01,490 --> 00:07:06,170
to regulate AI in a manner that balances innovation with ethical considerations.
