Facts

Sonko released from prison before Senegal election

On 14 March 2024 Ousmane Sonko and Bassirou Diomaye Faye were released from prison, ten days before the presidential election Faye won. Their release followed an amnesty law passed by the National Assembly on 6 March and promulgated on 13 March, which human rights advocates criticised for also shielding officials linked to protesters' deaths.

• Ousmane Sonko

EU Artificial Intelligence Act adopted

On 13 March 2024, in Strasbourg, the European Parliament adopted the Artificial Intelligence Act, the ordinary-legislative-procedure text later published as Regulation (EU) 2024/1689. The Parliament press service recorded the plenary result as 523 votes in favour, 46 against and 49 abstentions (https://www.europarl.europa.eu/news/en/press-room/20240308IPR19015/artificial-intelligence-act-meps-adopt-landmark-law). Reuters reported the same vote the same day from Strasbourg (https://www.reuters.com/technology/eu-lawmakers-endorse-political-deal-artificial-intelligence-rules-2024-03-13/). That March vote was not yet the last formal step. The Council of the European Union gave its final green light on 21 May 2024 (https://www.consilium.europa.eu/en/press/press-releases/2024/05/21/artificial-intelligence-ai-act-council-gives-final-green-light-to-the-first-worldwide-rules-on-ai/). The signed instrument is dated 13 June 2024. The Official Journal published Regulation (EU) 2024/1689 on 12 July 2024 (https://eur-lex.europa.eu/eli/reg/2024/1689/oj). Article 113 provides that the regulation enters into force on the twentieth day following publication: 1 August 2024. The Commission recorded that entry-into-force date on 1 August 2024 (https://commission.europa.eu/news-and-media/news/ai-act-enters-force-2024-08-01en). The Fact recorded here is the 2024 adoption of that regulation as Union law. It is not a later national implementing statute, not a first enforcement fine, not a court judgment applying the Act, and not a measured change in model deployment. Staggered application dates inside the Act are part of the instrument; they are not separate Facts unless a later dated act of application or enforcement is recorded on its own page. The Commission that proposed the text in April 2021 was headed by Ursula von der Leyen. Digital-market and competition work in that College sat with Executive Vice-President Margrethe Vestager. Internal Market Commissioner Thierry Breton was the Commission member who publicly welcomed the March 2024 Parliament vote. Those names identify the College that carried the file. They do not convert the Fact into a biography. The Commission presented the proposal for a regulation laying down harmonised rules on artificial intelligence on 21 April 2021 (procedure 2021/0106(COD)). The file sat in the “A Europe fit for the digital age” programme. The Council adopted its general approach on 6 December 2022. Interinstitutional negotiations closed with a provisional political agreement on 9 December 2023 after a multi-day trilogue. Parliament’s Internal Market (IMCO) and Civil Liberties (LIBE) committees were the joint lead committees; co-rapporteurs Brando Benifei and Dragos Tudorache presented the March 2024 plenary vote. The 2024 instrument is a regulation, not a directive. It is directly applicable in the Member States on the dates the Act itself sets. It does not, by itself, create a new United Nations treaty, a US statute, or a Chinese administrative measure. Parallel non-EU instruments from 2023–24 (an executive order in Washington, a UK safety summit, Chinese generative-AI measures) are neighbouring Facts if recorded separately. They are not this Fact. The Act uses a risk-based architecture that the legal text itself states: prohibited practices; high-risk systems with listed obligations; limited-risk transparency duties including labelling of certain synthetic content; and a lighter touch for systems treated as minimal risk. It also addresses general-purpose AI models, including additional duties for models classified as posing systemic risk. Those classifications are the Act’s own legal categories. This page does not invent a count of systems in each bucket, because the Official Journal text does not publish such a census. Application is staggered, again as the instrument states. Prohibited-practice rules apply six months after entry into force. Governance and general-purpose-AI duties apply at twelve months. Most remaining obligations apply at twenty-four months. Some high-risk product-safety alignments run to thirty-six months. The Commission’s 1 August 2024 note restates those horizons. Entry into force is not the same event as full application. The Act also names an EU AI Office inside the Commission to support implementation. Establishing that office as a staffed unit is a later dated administrative fact if recorded on its own page. Mentioning the Office here identifies an institution the regulation contemplates; it does not duplicate that later event. The primary legal record is Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence, published in the Official Journal on 12 July 2024 (https://eur-lex.europa.eu/eli/reg/2024/1689/oj). That text is the adopted instrument: title, date, ELI, and operative articles including the entry-into-force clause. The Parliament’s 13 March 2024 press release is the official contemporaneous record of the plenary vote, the 523–46–49 tally, and the co-rapporteurs’ statements (https://www.europarl.europa.eu/news/en/press-room/20240308IPR19015/artificial-intelligence-act-meps-adopt-landmark-law). The same page links the adopted text, the 12 March 2024 plenary debate, the roll-call, and procedure file 2021/0106(COD). The Council’s 21 May 2024 press release is the official record of the Council’s final adoption (https://www.consilium.europa.eu/en/press/press-releases/2024/05/21/artificial-intelligence-ai-act-council-gives-final-green-light-to-the-first-worldwide-rules-on-ai/). The Council policy page on the AI Act restates the December 2023 political agreement, the May 2024 adoption, and 1 August 2024 entry into force (https://www.consilium.europa.eu/en/policies/artificial-intelligence-act/). The Commission’s 1 August 2024 communication records entry into force and restates the risk-based structure and staged applicability (https://commission.europa.eu/news-and-media/news/ai-act-enters-force-2024-08-01en). Independent contemporaneous reporting: Reuters, 13 March 2024, “Europe one step away from landmark AI rules after lawmakers’ vote,” including the 523/46/49 tally and a quoted statement from Internal Market Commissioner Thierry Breton welcoming the Parliament vote (https://www.reuters.com/technology/eu-lawmakers-endorse-political-deal-artificial-intelligence-rules-2024-03-13/). Earlier Reuters items on 2 February 2024 (Member State endorsement of the political deal) and 13 February 2024 (committee ratification) are process reports, not the March plenary act. Accounts conflict on political meaning, not on whether the regulation was adopted. Industry groups described compliance cost and legal uncertainty. Civil-liberties groups described remaining law-enforcement exceptions for remote biometric identification as too wide. Those characterisations are attributed opinion. The Fact record stores the adoption. In Factrail terms this is a recorded action: the Union co-legislators adopted a regulation. The Commission proposed; Parliament and Council adopted; the Official Journal published; the Act entered into force on the date the text itself sets. Margrethe Vestager, Thierry Breton, and Ursula von der Leyen are named because they held the Commission posts that carried the file. Naming them does not assign a numeric contribution size. Size is not invented here. The attached driver is AI regulation intensity. The live impact path treats the 2024 adoption as a strengthening of that driver: a binding, directly applicable Union instrument now exists where a proposal had existed since 2021. That is a legal-status change, not a measured change in model capability, compute concentration, or disinformation volume. If later enforcement, codes of practice, or court cases alter those neighbouring drivers, each needs its own dated record. The Act’s stated maximum administrative fines (up to 35 million euros or 7 percent of worldwide annual turnover for listed infringements, with lower bands for other infringements) are clauses in the adopted text. They are not a 2024 enforcement event. No fine total is recorded here because no such completed enforcement census is the Fact. Related graph neighbours already on the site, if present, include the people named above and the driver named above. They are not extra copies of this Fact. Established: on 13 March 2024 the European Parliament adopted the AI Act by 523 votes to 46 with 49 abstentions; on 21 May 2024 the Council adopted it; the signed regulation is dated 13 June 2024; Official Journal publication was 12 July 2024 as Regulation (EU) 2024/1689; entry into force was 1 August 2024; application is staggered as the Act provides; the Commission College that proposed the file included von der Leyen, Vestager, and Breton. Not established: that the Act caused a measured reduction in harmful AI uses, a measured increase in compliance cost across the Union economy, a change in frontier-model release practice outside the Union, or a quantified movement in any welfare indicator. Those would be separate Facts, each with their own date, instrument, and evidence. Fine maxima in the legal text are not applied fines. What would change the picture: a later Union instrument repealing or materially amending Regulation (EU) 2024/1689; a Court of Justice judgment that rewrites the operative scope; a dated first wave of published enforcement decisions with identified addressees; or a Member State measure that the Court treats as incompatible with the regulation. Until such a record exists, the Fact remains the 2024 adoption.