EU Artificial Intelligence Act
Recent News
Deal is Final on EU AI Act, Sweeping AI Legislation to be Adopted (Dec. 7, 2023)
[Council press release]
European Union squares the circle on the world’s first AI rulebook, Luca Bertuzzi, Euractiv, Dec. 8, 2023
After a 36-hour negotiating marathon, EU policymakers reached a political agreement on what is set to become the global benchmark for regulating Artificial Intelligence.
The AI Act is a landmark bill to regulate Artificial Intelligence based on its capacity to cause harm. The file passed the finishing line of the legislative process as the European Commission, Council, and Parliament settled their differences in a so-called trilogue on Friday (8 December).
At the political meeting, which set a new record for interinstitutional negotiations, the main EU institutions had to go through an appealing list of 21 open issues. As Euractiv reported, the first part of the trilogue closed the parts on open source, foundation models and governance. . . .
Source: Euractiv
News Reports
- AP, Europe reaches deal on world's first comprehensive AI rules, Dec. 8, 2023
- BBC, AI: EU agrees landmark deal on regulation of artificial, Dec. 8, 2023
- Bloomberg, EU Strikes Deal to Regulate ChatGPT, Other AI in Landmark Act, Dec. 8, 2023
- CNN, European Union agrees to regulate potentially harmful effects of artificial intelligence, Dec. 8, 2023,
- The Guardian, EU agrees ‘historic’ deal with world’s first laws to regulate AI, Dec 8, 2023
- New York Times, E.U. Agrees on Landmark Artificial Intelligence Rules, Dec. 8, 2023
Background
Luca Bertuzzi, AI Act enters final phase of EU legislative process, Euractiv, June 14, 2023
[Key Points]
Foundation models & generative AI
- The EU lawmakers introduced a tiered approach for AI models that do not have a specific purpose, so-called General Purpose AI, with a stricter regime for foundation models, large language models on which other AI systems can be built.
- The top layer relates to generative AI like ChatGPT, for which the European Parliament wants to introduce mandatory labelling for AI-generated content and force the disclosure of training data covered by copyright.
Main changes
- The MEPs introduced several other significant changes to the text, starting with the definition of AI aligned with the one of the OECD.
- The list of prohibited practices was extended to subliminal techniques, biometric categorisation, predictive policing, internet-scrapped facial recognition databases, and emotion recognition software is forbidden in law enforcement, border management, workplace and education.
- An extra layer was added for AI applications to fall in the high-risk category, whilst the list of high-risk areas and use cases were made more precise and extended in law enforcement and migration control areas. Recommender systems of prominent social media were added as high-risk.
- The obligations of high-risk AI providers concerning risk management, data governance, and technical documentation were made more prescriptive. New requirements were introduced to conduct fundamental rights impact assessments and monitor environmental impact.
- An AI Office was established to support coordination on cross-border cases, albeit the critical responsibility of settling intra-authorities disputes was left to the Commission. Tudorache’s view is to upgrade the AI Office into an all-compassing digital agency eventually.
Upcoming negotiations
- MEPs will now enter interinstitutional negotiations with the EU Council of Ministers, representing European governments, and the European Commission in the so-called trilogues.
- The negotiations will intensify once Spain takes over the rotating presidency of the Council in July, as Madrid has made finishing up the AI law its top digital priority.
- The main points of contention are set to be on high-risk categories, fundamental rights, and foundation models. In contrast, issues such as governance, innovation and AI definition will likely be solved at the technical level.
Source: Euractiv
Final Text of Report and Amendments for the EU AI Act
Responses to 11 May LIBE-IMCO Vote
- AccessNow, Big wins, but gaps remain: European Parliament Committees vote to secure key rights protections in AI Act
- EDRi, EU Parliament sends a global message to protect human rights from AI
- European Center for Nonprofit Law, The wins, concerns and what's missing: ECNL's reaction as the AI Act is adopted in the European Parliament committees
Introduction
With the introduction of the Artificial Intelligence Act, the European Union aims to create a legal framework for AI to promote trust and excellence. The AI Act would establish a risk-based framework to regulate AI applications, products and services. The rule of thumb: the higher the risk, the stricter the rule. But the proposal also raises important questions about fundamental rights and whether to simply prohibit certain AI applications, such as social scoring and mass surveillance, as UNESCO has recently urged in the Recommendation on AI Ethics, endorsed by 193 countries.
Because of the significance of the proposed EU Act and the CAIDP’s goal to protect fundamental rights, democratic institutions and the rule of law, we have created this informational page to provide easy access to EU institutional documents, the relevant work of CAIDP and others, and to chart the important milestones as the proposal moves forward.
We welcome your suggestions for additions. Please email us.
Recent news
(reverse chronological order)
- Euractiv, AI Act moves ahead in EU Parliament with key committee vote (11 May 2023)
- Reuters, EU lawmakers' committees agree tougher draft AI rules (May 11, 2023)
- Fox News, Here is how Europe is pushing to regulate artificial intelligence as ChatGPT rapidly emerges (May 11, 2023)
- POLITICO, Facial-recognition ban gets lawmakers’ backing in AI Act vote: The issue of biometric identification had been hotly debated by lawmakers (May 11, 2023)
- EU's main goal is to protect the continent against any AI-related threats
-
Euractiv, AI Act: EU Parliament fine-tunes text ahead of key committee vote (May 8, 2023)
- Euractiv, MEPs Seal the Deal on Artificial Act (Apr. 27, 2023)
- Euractiv, AI Act: MEPs close in on rules for general purpose AI, foundation models (20 April 2023)
- Euractiv, AI Act: EU Parliament walking fine line on banned practice (18 April 2023)
- Euractiv, Leading EU lawmakers propose obligations for General Purpose AI (14 March 2023)
- Euractiv, EU lawmakers set to settle on OECD definition for Artificial Intelligence (7 March 2023)
- Euractiv, AI Act: MEPs extend ban on social scoring, reduce AI Office role (28 February 2023)
- Euractiv, AI Act: EU Parliament’s crunch time on high-risk categorisation, prohibited practices (7 February 2023)
- Euractiv, AI Act: co-rapporteurs seek closing high-risk classification, sandboxes (26 January 2023)
- Euractiv, AI Act: MEPs want fundamental rights assessments, obligations for high-risk users (10 January 2023)
- Euractiv, Germany could become MEPs’ ally in AI Act negotiations (9 January 2023)
- Euractiv, Leading MEPs exclude general-purpose AI from high-risk categories – for now (12 December 2022)
- Euractiv, EU countries adopt a common position on Artificial Intelligence rulebook (6 December 2022)
- Euractiv, Leading MEPs tackle enforcement in AI regulation (22 November 2022)
- Euractiv, Artificial Intelligence definition, governance on MEPs’ menu (8 November 2022)
- Euractiv, Leading MEPs push for European ‘office’ to enforce the EU’s AI rulebook (4 November 2022)
- Euractiv, AI Act: Czech EU presidency makes final tweaks ahead of ambassadors’ approval (3 November 2022)
- Euractiv, AI Act: Leadings MEPs want to expand Commission’s revision powers (25 October 2022)
- Euractiv, The US unofficial position on upcoming EU Artificial Intelligence rules (24 October 2022)
- Euractiv, EU Council nears common position on AI Act in semi-final text (19 October 2022)
- Euractiv, AI Act: Czech Presidency puts forward narrower classification of high-risk systems (16 September 2022)
- Euractiv, Leading MEPs seek broad consensus on regulatory sandboxes in AI Act (2 September 2022)
- Euractiv, Leading MEPs tackle obligations for high-risk AI systems in new compromise (29 August 2022)
- The New Statesman, Inside Europe’s fight for ethical AI: France wants to expand the scope of the EU's groundbreaking new AI Act. Do its plans go far enough? (14 June 2022)
- Euractiv, AI Act: leading MEPs seek first compromise on administrative procedures, standards (July 8, 2022)
- Euractiv, AI regulation filled with thousands of amendments in the European Parliament (2 June 2022)
- EPRS, Regulatory divergences in the draft AI act: Differences in public and private sector obligations (May 2022)
- Euractiv, AI standards set for joint drafting among European standardisation bodies (30 May 2022)
- MIT Technology Review, A quick guide to the most important AI law you’ve never heard of (13 May 2022)
- European Parliament, Artificial intelligence: MEPs want the EU to be a global standard-setter (4 May 2022)
- AI Decoded, A Dutch algorithm scandal serves a warning to Europe — The AI Act won’t save us (30 March 2022)
- AI Decoded, The Tech Policy Fight of the Decade Begins (23 March 2021)
- Euractiv, EU lawmakers adopt recommendations on Artificial Intelligence (23 March 2022)
- European Parliament, Artificial Intelligence: MEPs and experts discussed the parameters of law proposal (21 March 2022)
- AI Decoded, [EU AIA] Ban Debate Heats Up (16 March 2022)
- AI Decoded, Parliament Kickstarts Debate (9 March 2022)
- Law Society Gazette Ireland, Call for ban on use of AI systems in law enforcement (3 March 2022)
- AI Decoded, Meet Parliament’s second AI man — France on the AI Act (19 January 2022)
- Techcrunch, Europe’s AI Act falls far short on protecting fundamental rights, civil society groups warn (30 November 2021)
Recent Events
22-03-2022 9:00-9:30, 13:45-15:15
"On 22 March (9:00-9:30 and 13:45-15:15) the AIDA Special Committee will vote on its report on artificial intelligence in a digital age (2020/2266(INI). Within the deadline in December last year, 1384 amendments were tabled, and the rapporteur and shadow rapporteurs held a large number of technical and shadows meetings to work on compromises.
"Throughout the process, via numerous hearings and workshops organised during the committee mandate of 18 months, the members and staff gathered expertise and insights on various aspects of AI. This work fed into the committee's final report, which aims to establish an AI Roadmap up to 2030. AIDA rapporteur: Axel Voss (EPP/PPE, Germany).
At this meeting the AIDA Chair and Members of the AIDA committee will also conclude the work of the committee."
The Artificial Intelligence Act
"The IMCO & LIBE committees will hold a joint hearing on the 21 March on the proposal for an Artificial Intelligence Act. The aim of the hearing is to address main issues concerning the Artificial Intelligence (AI) legislative proposal put forward by the Commission on 21 April 2021 and help preparing the Parliament's position.
"This piece of legislation, which is considered a main pillar of the EU digital single market strategy, will set out horizontal rules for the development, modification and use of AI-driven products, services and systems within the territory of the EU, applicable to all industries. The AI act aims to ensure the proper functioning of the internal market and the free movement of AI-based goods and services cross-border and to codify the high standards of the EU trustworthy AI paradigm, which requires AI to be legally, ethically and technically robust, while respecting democratic values and fundamental rights."
EU Institutional Documents on the AIA
EUROPEAN COMMISSION
- European Commission, Artificial Intelligence Act (COM/2021/206 final)
- European Commission, White Paper on Artificial Intelligence: a European approach to excellence and trust, ("EU White Paper") (19 February 2020)
- European Commission, A European Approach to Artificial Intelligence
- European Commission, Legislative Train Schedule (background information and visualization of status of EU AI Act)
COUNCIL OF THE EU
- Council of the EU, Artificial Intelligence Act: Council calls for promoting safe AI that respects fundamental rights (6 Dec. 2022)
- Council of the EU, Final Presidential Compromise Act (3 Nov. 2022)
- Council of the EU, Third Presidency Compromise Text (2021/0106(COD)) (23 Sept. 2022) (Czech Republic)
- Council of the EU, Recommendation for a COUNCIL DECISION authorising the opening of negotiations on behalf of the European Union for a Council of Europe convention on artificial intelligence, human rights, democracy and the rule of law (18 Aug. 2022)
- Council of the EU, Presidency Compromise Text (2021/0106(COD)) (29 Nov. 2021) (Slovenia)
- Council of the EU, [TEXT]
EUROPEAN PARLIAMENT
- Special Committee on Artificial Intelligence in a Digital Age, AIDA Draft Opinion (11 February 2022)
- Committee on Industry, Research and Energy, ITRE Draft Opinion (3 March 2022). pdf
- Committee on Culture and Education, CULT Draft Opinion (21 February 2021)
- Committee on Legal Affairs, JURI Draft Opinion (2 March 2022)
- "Digitizing Europe" (Documents and Timelines: The Artificial Intelligence Act)
EUROPEAN PARLIAMENT RESEARCH SERVICES
CAIDP and the EU AI Act
Brando Benifei at AI and Democracy conference, Harvard University, Dec. 1, 2023
CAIDP Meets with MEP Juan Fernando López Aguilar and Delegation from LIBE Committee of the European Parliament, Washington, DC, May 26, 2022
CAIDP meets with MEP Dragoș Tudorache and Members of the European Parliament AIDA Committee, Washington, DC, 4 November 2021
CAIDP Statements on the EU AI Act
- CAIDP Statement on the Council General Approach (13 February 2023)
We urge the Swedish Presidency to commit to completing the trilogue in the first half of 2023. In the absence of a legal framework, AI systems will be deployed without necessary safeguards, putting at risk public safety and health, and fundamental rights. . . .
CAIDP recommends that the Council of the European Union, the European Parliament, and the European Commission adopt the following changes ahead of / during the trilogue process. These changes would ensure the Proposal is in alignment with the Council’s stated goals, further protect fundamental rights and Union values, and bring consistency to the Approach. . . .
Prohibit Pseudoscientific and Discriminatory AI Systems
● Require scientific validity for AI systems
● Ban predictive policing
● Ban emotion recognition systems
● Ban biometric categorisation systems
Apply bans to both public and private entities
Safeguard Fundamental Rights
- Remove the broad exclusions for law enforcement
- Remove the exclusions for ex ante systems
- Remove the national security exclusion
- Correct the Unequal Protection of Asylum Seekers and Refugees
Ensure Transparency and Accountability
- Mandate ex ante impact assessments
- Record serious incidents
- Require private users to registers
- Mandate independent, third-party auditing
- Regulate General Purpose AI (GPAI) systems
- Establish obligation to terminate A systems no longer under human control
Protect Societal Interests
- Protect the environment
- Safeguard Disability Rights
- Adopt UNESCO Recommandation on AI Ethics
CAIDP congratulates LIBE – IMCO for the amendment to explicitly protect children’s rights, provide group-based complaint and remedy mechanisms, and for additional protections against misinformation – as these were three of the five recommendations we had initially proposed [to address societal risks]. We still note that there are further risks to the society for the Act to address to protect the environment and disability rights.
- CAIDP Statement on Proposed EU AI Regulation (28 July 2021) (excerpts)
-
- CAIDP recommends that all prohibited use cases for AI systems should apply equally to both private and public entities. If a system is a risk to the fundamental rights of an individual and is detrimental to society, then it does not matter which entity uses it and it should be prohibited in both remits.
- CAIDP recommends that the right to accessibility should apply to everyone, as various barriers may hinder full and effective participation in society on an equal basis with others.
- CAIDP recommends a ban on biometric recognition systems used for mass surveillance purposes. . . . This ban should not only be limited to facial recognition systems, but also new /emerging forms of biometric recognition analyses such as gait, voice, etc
- CAIDP recommends a ban on any score-based profiling of individuals by private companies that is not fully compliant with all legal obligations prior to deployment and for which an algorithmic impact assessment has not been conducted ex ante.
- CAIDP recommends a ban on the use of pseudoscientific AI systems to detect emotional state of a natural person, as well as use of what is described as “predictive policing.”
- CAIDP recommends that asylum seekers and refugee rights be protected on an equal basis and that these populations not become test beds or experimentation for emerging technologies, as has been the case with border control systems requiring biometric identifiers, such as fingerprints and facial images, in the recent years.
- CAIDP recommends that the Proposal explicitly include ‘examination of bias in model design and selection of performance metrics.’
- CAIDP recommends conformity assessments for high-risk systems be conducted by certified independent third parties who shall not have any conflict of interest.
- CAIDP repeats the recommendation for a ban online categorization and scoring of individuals using biometric features (facial features, voice, DNA data) by both public and private entities.
- CAIDP Statement to the AIDA Committee on Artificial Intelligence and the Digital Age Report (16 December 2021)
We applaud Europe’s goal to establish comprehensive regulations for the development, design, and deployment of Artificial Intelligence . . . However, we are concerned that the Report in its current state would both amplify problems previously identified in the proposed EU AI Act and introduce new difficulties. And we fundamentally disagree with an assessment that a regulatory framework that supports human-centric and trustworthy AI is at odds with innovation. CAIDP sets out below a dozen recommendations in response to the Report.
- CAIDP Statement to European Commission, European Parliament, and European Council regarding the Draft EU AI Regulation (20 April 2021)
We write to express support for the draft EU AI Regulation and to offer technical advice as amendments to the text are considered. This initiative may be the single most important legal framework for the digital economy to ensure the protection of fundamental rights.
- CAIDP Statement on "External Dimensions of AI" for European Parliament (1 March 2021).
We see particular urgency in the need to safeguard democratic values, including fairness, accountability, and transparency, in a global framework with legal force. Technology advances at a quick pace. Policymaking is necessarily slower and more deliberative. There is a real risk that systems for mass surveillance and opaque processing will take hold before there is an opportunity for debate and assessment
CAIDP - 5 Key Points about the EU AIA
Perspectives of Members of CAIDP Global Academic Network (GAN)
(reverse chronological order)
- Katja Langenbucher, AI credit scoring and evaluation of creditworthiness - a test case for the for the EU proposal for an AI Act, ECB Legal Conference 2021 (April 2022)
- Stuart Russell, Statement for the IMCO-LIBE Hearing on the AI Act (22 March 2022)
- Niovi Vavoula, Unpacking the EU Proposal for an AI Act: Implications for Systems Used in the Context of Migration, Asylum, and Border Control Management, Turkish Policy Quarterly (4 March 2022)
- Alex Engler, The EU and U.S. are starting to align on AI regulation, Brookings (1 February 2022)
- Leyla Keser, A Paradigm Shift From “Ethics” To “Law” Regarding AI Regulations, International Association of Law Libraries (27 April 2021)
- Irena Barkane, Questioning the EU proposal for an Artificial Intelligence Act: The need for prohibitions and a stricter approach to biometric surveillance, Information Policy, vol. 27, no. 2, pp. 147-162, 2022
AI and Democratic Values Index
Relevant Provisions:
- Findings
- Recommendations
- New Recommendations (2022)
- The European Commission
- The European Parliament
- Committees - AIDA, IMCO, LIBE
- The Two Councils
- The Metrics
- OECD AI Principles
- Universal Guidelines for AI
- UNESCO Recommendation of AI Ethics
- GPA Declaration on Ethics and Data Protection in AI
- GPA Resolution on AI and Accountability
Reports of the European Law Institute on AI
The European Law Institute "seeks to initiate, conduct and facilitate research, to make recommendations, and to provide practical guidance in the field of European legal development. . . . It will study and stimulate European legal development in a global context. That should be taken to include, but by no means be limited to, the development of European law by the European Union and the Council of Europe.”
Reports and Recommendations
- AccessNow, The EU AI Act proposal: a timeline (14 July 2022)
- AccessNow and 52 Organizations, Statement on Need to Prohibit Remote Biometric Identification (10 May 2022)
In Europe and across the world, the use of remote biometric identification (RBI) systems such as facial recognition, in our publicly accessible spaces, represents one of the greatest threats to fundamental rights and democracy that we have ever seen.
The remote use of such systems destroys the possibility of anonymity in public, and undermines the essence of our rights to privacy and data protection, the right to freedom of expression, rights to free assembly and association (leading to the criminalisation of protest and causing a chilling effect), and rights to equality and non-discrimination.
- BEUC, Regulating AI to Protect the Consumer: Position Paper on the AI Act (7 Oct. 2021)
the AI Act proposal requires substantial improvements to guarantee consumers have the protections they need and can trust AI to respect their rights and freedoms.
- BEUC, Artificial Intelligence: what consumers say: Findings and policy recommendations of a multi-country survey on AI (Sept. 2020)
While they see benefits, consumershave low trust in AI and have concerns about the abuse of personal data and the use of AI to manipulate decisions
- BEUC, AI Rights for Consumers (Oct. 23, 2019)
Legislators must make sure that AI products and services are safe and that risks – including discrimination, loss of privacy, loss of autonomy, and lack of transparency – are avoided.
- ECNL, How can the EU include stakeholders to ensure effective implementation of the EU AI Act? (16 March 2022)
The first set of ECNL's proposals on the EU AI Act amendments focuses on the governance structure and supervisory tasks.
- EDRi, Civil society calls on the EU to put fundamental rights first in the AI Act (30 November 2021)
Today, 30 November 2021, European Digital Rights (EDRi) and 119 civil society organisations launched a collective statement to call for an Artificial Intelligence Act (AIA) which foregrounds fundamental rights.
- ECNL, Evaluating the risk of AI systems to human rights from a tier-based approach (23 March 2021)
The European Commission should consider risks of AI systems within a rights-based framework - as risks they pose to human rights, rule of law and democracy.
- ICCL, European Parliament embraces ICCL recommendations on the Artificial Intelligence Act (5 May 2022)
Most of our recommendations on enforcement have been embraced in the report . . . These include strengthening the enforcement at Union level, giving individuals the right to complain and the right to judicial remedy. The right to lodge a complaint to a supervisory authority has also been included in the JURI report, albeit limited to the infringement of data protection. IMCO-LIBE report rightly eliminated this limitation.