AI Act: Key Provisions of the First Major Law Regulating AI

This presentation outlines the significant aspects of the AI Act, the first major legislation regulating artificial intelligence (AI) in the European Union, which officially came into force on August 1, 2024. The AI Act is structured into 13 chapters and 102 articles, emphasizing a risk-based approach to AI regulation, classifying AI systems into minimal, high, and unacceptable risk categories.

Key points include:

- Minimal Risk: AI systems like spam filters and video games are exempt from strict requirements.
- High Risk: Systems such as those used in medical devices and credit decisions face stringent regulations, including human oversight and regular testing.
- Unacceptable Risk: Certain AI systems, like social rating systems, are completely banned.
The presentation also highlights the significant implications for major tech companies, with potential fines up to 7% of global annual revenue for non-compliance. The AI Act sets a precedent for future AI regulations globally, with an AI Office established to oversee implementation and promote best practices.

1. AI
ACT
KEY PROVISIONS
the First Major Law Regulating AI
25.08.2024

2. I am currently pursuing a Master's degree in Information Technology in
Wellington.
I AM: IVAN KOVALEV
DATA ANALYST
A LAWYER SPECIALIZING IN LAW
SPECIALIST IN THE FIELD OF CREDIT RISK MANAGEMENT

3. *"The law is strict, but it is the law" (Latin)
"DURA LEX,
SED LEX."
THE PHILOSOPHY OF ANCIENT ROMAN LAW
"The law is not just a
support for authority.. . .
. . . but a safeguard
against anarchy."
Plato (~427-347 B.C.)

4. 'LAWS MAY SOMETIMES BE OUT OF STEP WITH THEIR TIME,
BUT THEY ARE ALWAYS BETTER THAN CHAOS.'
POINT OF VIEW:
- Edmund Burke (1729–1797), statesman, philosopher and orator

5. In April 2021
the European Commission introduced a draft law on
the regulation of artificial intelligence (AI Act).
HISTORY OF CREATION
On August 1, 2024
the AI Act was adopted by the European Union and
officially entered into force.

6. UNESCO Recommendations on the Ethical Aspects of AI
Value-based approach and principles for the operation of a universal AI
model.
International Standard ISO
"Artificial Intelligence—Ethical and Social Issues—General Provisions"
[ISO 2022].
Principles for Responsible AI
from the Organisation for Economic Co-operation and Development
(OECD) ratified by several countries.
WHY IS
AI ACT
1ST
Framework Recommendations and Initiatives of the U.S.
The framework for managing AI-related risks released in 2021.

7. MAJOR LAW CONSEQUENCES OF
NON-COMPLIANCE
FOUNDATION OF LEGAL
REGULATION
AI ACT: SIGNIFICANCE FACTORS

8. Chapters.
I: General provisions
II: Categories of artificial intelligence
III: Requirements for high-risk AI systems
IV: Conformity assessment procedures
V: General AI ModelsChapter VI: Information transparency requirements
VII: Governance at the Union level
VIII: EU database for high-risk AI systems
IX: Post-market monitoring, information exchange and supervision
X: Sanctions and fines
XI: Final provisions
XII: Applications
XIII: Final and transitional provisions
AI ACT: STRUCTURE
144 pages
13 chapters
102 articles

9. AI Act applies a risk-based approach to the
regulation of artificial intelligence, which means
classifying AI systems based on the level of risk
they may pose to society and individuals.
MAIN PROVISIONS:
RISK-BASED
APPROACH
Minimal risk High risk Unacceptable risk

10. CATEGORY:
MINIMAL RISK
They are exempt from the mandatory requirements of the AI Act, but
companies can voluntarily adopt additional codes of conduct
EXAMPLES:
Spam filters
AI systems in video games.

11. They must clearly inform users that they are interacting with AI, as well as
provide labeling of AI-generated content.
SUBCATEGORY:
SPECIFIC RISK
EXAMPLES:
Chatbots
Сontent generation systems
REQUIREMENTS:

12. Strict requirements, including risk assessment and implementation of
mitigation measures. Ensuring the high quality of the data used. Mandatory
presence of human control. Regular testing.
CATEGORY:
HIGH RISK
EXAMPLES:
AI systems used in medical devices
Credit decision-making systems
Recruitment systems
REQUIREMENTS:

13. A complete ban on the use of such AI systems
CATEGORY:
UNACCEPTABLE RISK
EXAMPLES:
Social rating systems
Systems for predicting crimes based on people's
profiles
REQUIREMENTS:

14. GENERATIVE AI
THE MAIN PROVISIONS OF THE AI ACT FOR
Copyright protection
Generative AI systems must take copyright and intellectual property into
account when creating content. This applies, for example, in cases where AI
creates images or text based on copyrighted works.

15. Companies: Microsoft, Google, Meta and other major players
IMPACT
ON LARGE TECHNOLOGY COMPANIES
COMPANIES:
BASIC REQUIREMENTS:
FINES:
Compliance with transparency, data protection, and ensuring the
security of AI systems.
up to EUR 35,000,000 or up to 7% of the company's global annual
revenue for violating the law.

16. A new body created to oversee the implementation of the
provisions of the AI Act.
THE FUTURE OF AI REGULATION
AI Office:
Code of Practice:
AI Office:
Development and implementation of voluntary codes of
conduct and recommendations on the use of AI
The AI Act will definitely become an example for other countries in
improving AI regulation.

17. LET'S SUMMARIZE
The full text of the Law:
“Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024
laying down harmonised rules on artificial intelligence and amending Regulations (EC) No
300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU)
2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial
Intelligence Act)”
http://data.europa.eu/eli/reg/2024/1689/oj
Other sources used:
“AI Act enters into force”
https://commission.europa.eu/news/ai-act-enters-force-2024-08-01_en
“In world first, EU's sweeping AI law enters into force”
https://techxplore.com/news/2024-08-world-eu-ai-law.html

18. THANK YOU!