Timeline:

• On 21 April 2021, the EU Commission published its initial draft of the AI Act (see pdf)
• On 28 September 2022, a proposed AI liability Directive (AILD) was adopted
• 6 December 2022: The Council has adopted its common position ("general approach") on the AI Act
• On 14 June 2023, the European Parliament published its final position on the Artificial Intelligence Act, with substantial amendments to the Commission's text, including an updated definition of AI systems (due to the arrival and success of generative AI), an altered list of AI systems prohibited in the EU (in particular biometric identification systems, social scoring and "emotional recognition"), and a clearer definition of risks.
• June 2023: EU lawmakers have started the negotiations to finalise the new legislation
• EU Parliament committees will continue to work on the AI Liability Directive and the Revised Product Liability Directive, which will also have great impact on AI developments in the upcoming years
• December 8th, 2023: the Trilogue (European Commission, European Parliament, and Council of the EU) agreed in Brussels on a final version of the legislation via the so-called Trilogue negotiation process
• Press release on the Artificial Intelligence Act: deal on comprehensive rules for trustworthy AI on the banned applications, law enforcement exemptions, obligations for high-risk systems, guardrails for general-purpose AI (GPAI) systems, Measures to support innovation and SMEs, ...
• The Parliament’s Internal Market and Civil Liberties committees agreed on February 2nd, 2024 unanimously (after some armwrestling with France and Germany) to publish the final draft text. You can find the final draft, including all Annexes and Recitals, at this link.
• The 280-page 'Final draft' text is a bit of a nightmare to find your way around. This is why we love the AI Act Explorer web site, which makes the AI Act text fully navigable and searchable via a pretty web interface.
• This draft final text is now forwarded to a plenary vote by the European Parliament mid-April 2024. After a two-year implementation period, the AI Act will finally be enforced
• Sidenote: To understand Europe's thorough but complex and time-consuming decision process, check out this page on standards setting by the EU

The best sites to learn more about the AI Act:

• Future of Life Institute's website on The Artificial Intelligence Act. FLI is actually an independent non-profit, but many people think that their website is the official EU AI Act website.
• FLI also has an EU AI Act Compliance Checker
• The EU AI Act flowchart by Burges Salmon
• The official website with all the links to the different European digital initiatives, can be found at "A European Approach to AI"
• #DigitalEU's Youtube channel covers not just the upcoming European AI legislations efforts, but also the European Identity Wallet (#EUDI), Europe's (re)skilling programme, and many other European digital initiatives
• The European Commission's AI Watch website, in particularlty their projects and publications on trustworthy AI
• On twitter, you should follow Robotics & AI - EU
• One of the best explanations of the EU AI Act can be found in the Expert Explainer: The EU AI Act Proposal" at the Ada Lovelace Institute
• The European AI Alliance, a forum to establish an open policy dialogue on AI. Free membership, highly recommended.
• Oxford Commission on AI & Good Governance: Harmonising Artificial Intelligence on The Role of Standards in the EU AI Regulation (Dec. 2021)
• Citadel AI, a leader in trustworthy AI technology that works together with the British Standards Institution, published An Overview of AI Standardization and Regulation in 2023
• The gobal member-driven Responsible AI Institute (RAI Institute) took https://www.responsible.ai/post/eu-ai-act-explained">a good look at the EU AI Act
• We have a list of interesting people and news sources on the AI Act on X/Twitter

Recommended Background Reading on Fairness, Transparency and Explainability in Ethical and Trustworthy AI:

• 📔 A great resource about Trustworthy AI is Arnoud Engelfriet's “AI & Algorithms” book, which also explains how to use the ALTAI self-assessment tool
• The High-Level Expert Group on AI's Ethics Guidelines for Trustworthy Artificial Intelligence (2019, pre-AIAct)
• The Trade and Technology Council (TTC) published a document on U.S.-EU Joint Roadmap on Evaluation and Measurement Tools for Trustworthy AI and Risk Management to bring EU and U.S. approaches closer together
• The human rights association Access Now has several articles on the EU AI Act, a human rights impact assessment for AI and a 24/7 Digital Security Helpline
• The Belgian Knowledge Centre Data & Society and their vision paper on on how to ensure effective enforcement of the AI Act
• The U.S Government’s Accountability Office (GAO) published the Artificial Intelligence Accountability Framework for Federal Agencies and Other Entities in 2021 (focuses on accountability and responsible use of AI)
• For Lawyers: ITECHLAW (the IT Law Association) published "Responsible AI: A Global Policy Framework" back in 2019. 300 pages about accountability, transparency, explainability, safety, fair competition, privacy, intellectual property, ... from a legal point of view
• For auditors: ISACA's free white paper "Auditing Artificial Intelligence" (of course, this is based on COBIT)
• Also for auditors: The IIA's publications on Internal Audit's Role and Artificial Intelligence Auditing Framework (only for IIA users)
• For engineers: IEEE CertifAIEd is a certification program for assessing ethics of Autonomous Intelligent Systems (AIS). IEEE now also has a free Awareness Module on AI Ethics and Governance
• We wrote a LinkedIn article with many pointers to #XAI and trustworthy AI sites back in 2020
• The Open Data Institute's Data Ethics Maturity Model Data for benchmarking your approach to data ethics, based on the Data Ethics Canvas (including course and printable guides)
• The Asilomar AI Principles, 23 AI Governance guidelines for the research and development of artificial intelligence (AI), developed in 2017, still generally valuable
• The Dutch ECP | Platform voor de InformatieSamenleving has a downloadable Artificial Intelligence Impact Assessment checklist and code of conduct (pre-EUAIAct and pre-generativeAI, but still 94 useful pages
• ForHumanity (a non-profit public charity) Certified Auditors (FHCA) Code of Ethics
• The National Conference of State Legislatures released their report aimed at providing states with guidance on regulating AI (see linkedin post)
• The Association of AI Ethicists
• Google produced the People+AI Guidebook on Methods, best practices and examples for designing human-centered AI products" with a chapter on Explainability + Trust (pre-EUAIAct but useful as checklist)
• The Ethical OS Toolkit, combining a risk checklist, scenarios, strategies and use cases
• Google has a set of Explainable AI Tools and frameworks to help you understand and interpret predictions made by your machine learning models. Not sure if these tools are up-to-date with the latest LLM and generative AI developments.
• IBM Watson also has explainable artificial intelligence (XAI) processes and methods that allow human users to comprehend and trust the results and output created by machine learning algorithms. Not sure if these tools are up-to-date with the latest LLM and generative AI developments.
• AI Verify Foundation offers an AI governance testing framework and software toolkit that help industries be more transparent about their AI to build trust (based in Singapore)
• To read more about the impact of AI on our jobs and society, check out 80000hours' pretty harsh but wel-documented article (80000 is roughly the number of hours in your career)
• Microsoft has a set of Hands-on tools for building effective human-AI experiences, the HAX Toolkit for teams building user-facing AI products (but just a bit about social bias, nothing about fairness and transparency)

Knowledge sharing about the EU AI Act:
 

• We work together with national and international experts in the field of AI and law, and plan several sessions on how to prepare and be ready for this new regulation
• We invite tool vendors and consultants to contact us to investigate how we can work together
• If you are interested in a brief introductory meeting, do not hesitate to use calendly to book a meeting
• Join our dedicated low-volume AI Act newsletter list to stay informed about interesting announcements, articles, frameworks, reports, tools, ... that may be interesting to you