Polity Atlas is independent and free for everyone. Help keep it that way. Keep Polity Atlas independent. Support Polity Atlas →

Polity Atlas
← All authors

Polity Atlas editorial team

Editorial team

Briefings and research notes written and reviewed by the Polity Atlas editors.

Recent writing

2026-10-24EstablishedContesting automated decisions: explanation, human review and appealHow the GDPR, the CJEU's SCHUFA and Dun & Bradstreet Austria judgments, the EU AI Act, the Council of Europe AI convention, the UK's 2025 reforms and US rules give people a way to challenge automated decisions — and what evidence shows works in practice.2026-10-23HypothesisSignal: France's CESE convenes a citizens' convention on public finances on its own initiativeIn October 2026 France's Economic, Social and Environmental Council began recruiting 150 randomly selected citizens for a convention on public finances, due to start on 27 November 2026. It is the first convention the CESE has launched on its own initiative rather than at the executive's request.2026-10-23DisputedAutomation, AI and jobs: what the evidence shows and what is still forecastRobots have had measurable local effects on jobs and wages; generative AI raises productivity in specific tasks; administrative data to mid-2026 show no economy-wide displacement but fewer young hires in exposed occupations. Claims about future mass unemployment remain forecasts.2026-10-22EstablishedResource dividends and social wealth funds: lessons from Alaska, Norway, Iran and MongoliaAlaska pays every resident a dividend, Norway saves its oil wealth under a fiscal rule, Iran replaced subsidies with a universal transfer and Mongolia promised resource cash to all citizens. Dividends proved compatible with work, but they lasted only where payouts were tied to a rule-based draw from an accumulated fund.2026-10-21EstablishedUS guaranteed income pilots: what Stockton, OpenResearch, Cook County and Baby's First Years foundUnconditional monthly cash eased hardship and short-term stress in US pilots, but the largest randomised trial found a modest fall in work and fading wellbeing gains, and a trial of cash for mothers of newborns found no effects on child development or health at age four.2026-10-20EstablishedAutomated welfare enforcement compared: seven documented systemsA comparative study of SyRI, the Dutch childcare-benefit risk model, Michigan's MiDAS, Australia's Robodebt, Denmark's fraud-control models, the UK DWP advances model and Serbia's Social Card: what failed, who noticed, and which safeguards the evidence supports.2026-10-20EstablishedKenya's long-term basic income study: cash works, and how it is delivered mattersGiveDirectly's experiment in 295 Kenyan villages compares a 12-year basic income with a two-year stream and a lump sum. Two-year results show local business expansion, no fall in total work and lower depression; the lump sum matched the long-term stream on economic outcomes, while the short-term stream did less.2026-10-13EstablishedFederal and confederal designs compared: from the Articles of Confederation to RojavaSwitzerland, the United States before and after 1789, the Haudenosaunee Confederacy, the EU, north-eastern Syria and Ethiopia set against Riker, Stepan and Elazar. Confederations without revenue and with unanimity rules tend to be replaced; federations last when their safeguards are multiple; party-run ethnic federations are brittle.2026-10-06EstablishedDo people accept algorithmic decisions in government? What surveys and experiments showPeople often trust algorithmic advice but reject algorithms as sole decision-makers, especially for welfare and justice. Appeals to a human, meaningful human involvement, explanations of individual decisions and trust in the deploying agency raise acceptance; UK concern about welfare-eligibility AI rose from 44% to 59% between 2022 and 2024.2026-10-05EstablishedSwiss direct democracy by the numbers: initiatives, referendums, money and minoritiesOf 244 federal popular initiatives voted by September 2026, 26 passed; referendums defeat about two in five challenged laws. Fiscal referendums are associated with lower spending, but initiatives targeting minorities have passed and municipal naturalisation votes discriminated by origin until councils took over.2026-09-22EstablishedWhat makes territorial autonomy last? Åland to Hong KongA comparison of autonomy arrangements in Åland, South Tyrol, Greenland, Bougainville, Catalonia, Scotland and Hong Kong. Durable autonomies combine external anchors, entrenchment requiring the region's consent, clear fiscal terms and internal power-sharing; autonomy without enforceable guarantees can be withdrawn.2026-09-21EstablishedAlaska's 2026 Permanent Fund Dividend set at $1,200Alaska announced on 21 September 2026 that this year's Permanent Fund Dividend is $1,200, including a $200 energy relief payment, up from $1,000 in 2025. The amount, set each year in the state budget, has ranged from $1,000 to $3,284 since 2020.2026-09-15EstablishedAI is everywhere in OECD governments; evaluation is notThe OECD Digital Government Outlook 2026 reports AI use in at least one government function in 35 of 36 surveyed countries, but only 10 report any evaluation.2026-09-14EstablishedCitizens' assemblies after two decades: what happened to their recommendationsA review of British Columbia, Ontario, Ireland, France, the UK, Ostbelgien and the German Bundestag, plus OECD and academic data, finds that bounded questions with a pre-agreed decision route see high uptake, while broad remits and open-ended promises mostly produce filtered or shelved proposals.2026-08-25EstablishedAI as a deliberation aid: evidence from Polis to the Habermas MachineOpinion mapping, automated moderation and AI-drafted group statements have helped people find common ground in Taiwan, in Stanford deliberations and in DeepMind's 2024 Science experiments. The same tools shift opinions and narrow the range of ideas, so who controls and audits them matters more than whether they work.2026-08-24EstablishedSortition in practice: what permanent bodies chosen by lot have shown so farStanding sortition bodies in Ostbelgien, Brussels, Paris, Milan and Gdańsk show that permanence is feasible and cheap, but they remain small, mostly advisory and reversible, and almost none publishes data on self-selection, attrition or implementation. Proposals for a chamber by lot remain untested.2026-08-18EstablishedSpecial jurisdictions and charter cities: what the evidence showsA review of special economic zones, legal enclaves and charter cities from Shenzhen to Próspera. The average zone grows at the national pace; China's early zones and narrow legal enclaves did better; projects seeking deep autonomy have so far failed the test of domestic legitimacy.2026-08-14HypothesisSignal: Albania's AI 'minister' a year on, with its procurement system still on paperAlmost a year after Albania presented the chatbot Diella as a 'minister' to rid public tenders of corruption, the head of the Public Procurement Agency told BIRN in August 2026 that the AI procurement system was still in preparation, with only the terms of reference drafted.2026-07-28HypothesisSignal: Irish assembly recommendations are followed up unevenlyIreland's biodiversity assembly reports high uptake by the government's own count, while the Dublin assembly's mayor proposal was still 'under consideration' in July 2026, more than three years after its report.2026-07-27EstablishedEU AI Act timetable changes: high-risk rules move to 2027-2028Regulation (EU) 2026/1744 (the Digital Omnibus on AI) entered into force on 27 July 2026. Annex III high-risk obligations now apply from 2 December 2027 and Annex I product-embedded systems from 2 August 2028.2026-07-27HypothesisSignal: a network-state project signs a memorandum with a national governmentOn 27 July 2026 Kazakhstan's Ministry of Artificial Intelligence and Digital Development signed a memorandum with Balaji Srinivasan's Network School, which has since launched in Astana. It is cooperation with a host state, not recognition, and the government says no state money is involved.2026-07-07EstablishedDigital identity systems compared: Estonia, India, Kenya, the EU and SwitzerlandA comparison of five national digital identity systems on inclusion, exclusion, security and oversight. Systems built before their legal framework (Aadhaar, Huduma Namba) were reshaped by courts; those legislated or voted on first (eIDAS 2.0, the Swiss e-ID) are voluntary and data-minimising but still unproven at scale.2026-06-23EstablishedWhy some intentional communities endure: commitment, religion and adaptationKanter's commitment mechanisms and Sosis's costly-signalling evidence tested against the kibbutz, the Hutterites, Twin Oaks, Auroville and nineteenth-century communes. Communities last when they make membership meaningful, balance the cost of leaving and adapt their economic rules; survival often means becoming less radical.2026-06-22EstablishedParticipatory budgeting after 35 years: an evidence reviewLarge Brazilian studies link long-running participatory budgeting to more health and sanitation spending, lower infant mortality and higher tax collection. Elsewhere PB is small and its effects modest, and Porto Alegre's own decline after 2004 and suspension in 2017 show how much it depends on political commitment.2026-06-18TheorySignal: citizens' assemblies becoming a routine instrumentIreland has convened several citizens' assemblies since 2016; random-selection bodies are increasingly part of the ordinary toolkit rather than one-off experiments.2026-05-22EstablishedNew York City's participatory budgeting draws a record voteIn the City Council's fifteenth participatory budgeting cycle, nearly 130,000 New Yorkers in 22 Council districts voted from 11 to 19 April 2026 on how to allocate almost $25 million; the Council called it the highest vote since the programme began in 2011.2026-05-20HypothesisSignal: temporary 'pop-up cities' as a testbed for network communitiesAfter Zuzalu in 2023, several groups have organised temporary residential gatherings to trial community-first governance without claiming sovereignty.2026-05-19DisputedOrder without the state: what stateless and non-state orders can and cannot provideEvidence from Somalia after 1991, the Icelandic Commonwealth, the medieval law merchant, Zomia and self-governing commons. Non-state orders enforce contracts and manage commons within bounded communities, but protect outsiders poorly and struggle to restrain concentrated force.2026-05-19EstablishedDAO governance in practice: what the on-chain evidence showsA review of empirical studies of DAO voting finds low participation, extreme concentration of voting power, delegation that mostly serves large holders, and governance that can be bought, as in the 2022 Beanstalk attack. Legal wrappers and bicameral designs such as Optimism's address some problems but remain largely unevaluated.2026-04-27EstablishedLiquid democracy in practice: super-voters, low turnout and a theoretical warningFrom the Pirate Party's LiquidFeedback and the LiquidFriesland pilot to delegated voting in DAOs, delegation has produced a few super-voters with decisive weight and low, declining participation. Theory shows simple delegation cannot guarantee better decisions than direct voting.2026-03-10EstablishedLessons from the Dutch childcare-benefits scandalA parliamentary inquiry found 'unprecedented injustice' in how tax authorities treated families suspected of fraud; the cabinet resigned in January 2021.2026-02-02EstablishedWhat Finland's basic income experiment did — and did not — showUnconditional payments to unemployed people had small employment effects and improved wellbeing; the design cannot tell us what a universal scheme would do.2026-01-22EstablishedSouth Korea's AI Basic Act takes effectKorea's Framework Act on the Development of Artificial Intelligence and the Creation of a Foundation for Trust took effect on 22 January 2026. It defines 'high-impact' AI, including state decisions on eligibility for public services, requires labelling of generative AI output and sets fines of up to 30 million won.2026-01-05EstablishedHow to read epistemic statuses in Polity AtlasEvery claim is labelled established, theory, disputed, hypothesis or forecast — and separately as verified or not. Here is what the labels mean.2025-12-11EstablishedUS executive order moves against state AI lawsExecutive Order 14365 of 11 December 2025 creates a Justice Department task force to challenge state AI laws, asks Commerce to list 'onerous' ones and links some federal broadband funds to the outcome. It asks for a federal framework that would preempt state rules, with carve-outs including state use of AI.2025-11-23EstablishedFrench citizens' convention on children's time adopts 20 proposalsOn 23 November 2025 the 133 randomly selected members of France's Convention citoyenne sur les temps de l'enfant, organised by the CESE at the President's request, adopted a final report with 20 proposals, including a 9 a.m. school start for adolescents and a ban on social media before 15.2025-10-10EstablishedItaly's national AI law enters into forceLaw No. 132 of 23 September 2025, Italy's framework law on artificial intelligence, entered into force on 10 October 2025. It is designed to be applied alongside the EU AI Act and keeps final decisions in public administration and the courts with human officials.2025-09-29EstablishedCalifornia enacts a transparency law for frontier AI developersSigned on 29 September 2025, California's SB 53, the Transparency in Frontier Artificial Intelligence Act, requires large frontier developers to publish safety frameworks, report critical safety incidents to the state and protect whistleblowers; the Attorney General enforces civil penalties.2025-09-28EstablishedSwiss voters approve a state-issued e-ID by a narrow marginOn 28 September 2025 Swiss voters accepted the federal e-ID Act by 50.39% to 49.61%, four years after rejecting an earlier e-ID law that would have relied on private issuers.2025-09-26EstablishedUK announces a national digital ID, mandatory for right-to-work checksOn 26 September 2025 the UK government announced a free digital ID for people living in the UK, to be mandatory for Right to Work checks by the end of the Parliament. It would be held on people's phones and there would be no duty to carry it.2025-08-26EstablishedUN creates a scientific panel and a global dialogue on AI governanceOn 26 August 2025 the UN General Assembly adopted resolution 79/325, establishing a 40-member Independent International Scientific Panel on Artificial Intelligence and an annual Global Dialogue on AI Governance. Both are limited to non-military uses of AI.2025-08-02EstablishedEU rules for general-purpose AI models start to applyFrom 2 August 2025 providers of general-purpose AI models placed on the EU market must meet the AI Act's documentation and copyright duties, with extra safety duties for models with systemic risk. A voluntary Code of Practice published on 10 July 2025 offers one way to show compliance.2025-07-03EstablishedInter-American Court recognises Nature as a subject of rightsIn Advisory Opinion OC-32/25, adopted on 29 May 2025 and notified on 3 July 2025, the Inter-American Court of Human Rights states that recognising Nature as a subject of rights is compatible with the inter-American system.2025-06-04EstablishedJapan enacts a promotion-first AI law with no penaltiesJapan's Act on the Promotion of Research and Development and Utilisation of AI-Related Technologies (Act No. 53 of 2025) was promulgated on 4 June 2025. It sets up a Cabinet AI Strategy Headquarters and a national AI basic plan, tells ministries to use AI actively, and imposes no fines.2025-04-09EstablishedGerman basic-income pilot: little labour-market withdrawal, better wellbeingA three-year German study of €1,200 monthly unconditional payments reported in April 2025 no significant withdrawal from work and clear gains in mental health and life satisfaction, in a small, selected sample.2025-02-27EstablishedEU court: people are owed an intelligible explanation of automated decisionsIn Dun & Bradstreet Austria (C-203/22), decided on 27 February 2025, the Court of Justice held that a person subject to an automated credit assessment must be told which personal data were used and how, in a way that lets them understand and challenge the result. Handing over an algorithm is not enough.2025-02-26EstablishedICSID tribunal lets the Próspera arbitration against Honduras proceedOn 26 February 2025 the tribunal in Honduras Próspera Inc. v. Honduras rejected Honduras' objection that local remedies had not been exhausted. It is a ruling on admissibility, not on the merits.2025-02-02EstablishedEU AI Act: bans on 'unacceptable-risk' AI start to applyFrom 2 February 2025 the prohibited practices in Article 5 of the EU AI Act, and its AI-literacy duty, apply. Social scoring, untargeted scraping of facial images and most emotion recognition at work and in education are banned; fines for breaches became applicable in August 2025.2024-12-17EstablishedUK makes algorithm transparency records mandatory for central governmentThe UK's Algorithmic Transparency Recording Standard was mandated across government departments in February 2024; a scope and exemptions policy published on 17 December 2024 defines which bodies and tools must publish records.2024-10-01EstablishedHonduras' Supreme Court rules ZEDE framework unconstitutionalThe 2024 ruling removes the legal basis on which Próspera and other ZEDEs were created; their future now runs through politics and international arbitration.