2026-02-252025/26:TU8 · p.7TUTransport & infrastructure
Safe and reliable AI
Safe and reliable AI: the EU AI Act as the framework, without a new parliamentary initiative
What's at stake: The decision affects how well risks such as discrimination, privacy violations and misuse of AI are managed as adoption accelerates, particularly for people affected by AI-assisted decisions in public-sector activities.
Case texts and AI reasoning are AI-translated from the Swedish originals; the Swedish text is authoritative.
What the vote decides
The question is whether the EU's new AI Act, together with ongoing agency assignments, is enough to manage the risks of AI, or whether parliament should require stronger measures, such as mandatory impact assessments before AI is used in public-sector activities.
Yes Yes — the committee proposal
The committee welcomes the EU AI Act as a legal framework for safe, rights-respecting AI systems, is aware of the risks but points to ongoing agency assignments on issues such as discrimination risk and the extensive work under way within government, agencies and industry, and currently sees no reason for its own initiative.
No No — the reservations
Reservation 14 (V)
The V party points out that AI can bring unwanted consequences such as biased or faulty data, lack of transparency, discrimination, reduced trust and harm to individuals or to how democracy functions; it demands that all public-sector use of AI be preceded by careful impact assessments and that the government ensure municipalities, regions and agencies get the support needed to manage these risks.
Summarised from the committee report. Party labels in the summary come from the report, not from the vote.
What do the votes mean here?
Yes The committee proposal: rejecting the motions (keeping things as they are).
No Backing the counter-proposal: reservation 14 by V.
Abstain The party marks a position of its own that is not one of the two final alternatives.
What each plan points to — and how the party voted
For each party: the vote its own election manifesto and party programme point to in this case, next to how the party actually voted. A plan can only point to Yes or No; a party can also abstain.
| Plan | Vote | Confidence | Evidence | ||
|---|---|---|---|---|---|
| V | No | No 0/18/0/4 | follows the plan | confidence: medium | not addressed |
| S | No | Abstain 0/0/93/13 | departs from the plan | confidence: medium | not addressed |
| MP | No | Abstain 0/0/16/2 | departs from the plan | confidence: high | derived |
| C | No | Abstain 0/0/21/3 | departs from the plan | confidence: medium | not addressed |
| L | No | Yes 13/0/0/3 | departs from the plan | confidence: high | derived |
| KD | No | Yes 17/0/0/2 | departs from the plan | confidence: medium | not addressed |
| M | No | Yes 60/0/0/8 | departs from the plan | confidence: low | not addressed |
| SD | No | Yes 62/0/0/9 | departs from the plan | confidence: low | derived |
Committee proposal
The Riksdag rejects the motions 2025/26:311 by Rashid Farivar (SD) demands 7 and 8, 2025/26:1774 by Jan Ericson et al. (M), 2025/26:1815 by Aylin Nouri and Eva Lindh (both S), 2025/26:2370 by Malin Östh et al. (V) demand 11, 2025/26:2832 by Niels Paarup-Petersen et al. (C) demand 6, 2025/26:3137 by Emma Ahlström Köster (M), 2025/26:3320 by Leila Ali Elmi et al. (MP) demands 1-7, 2025/26:3405 by Emma Berginger et al. (MP) demand 11, 2025/26:3425 by Jan Riise et al. (MP) demands 1-3 and 6-8, 2025/26:3623 by Morgan Johansson et al. (S) demand 69, and 2025/26:3652 by Aylin Nouri et al. (S) demand 12.
Original (Swedish)
Riksdagen avslår motionerna 2025/26:311 av Rashid Farivar (SD) yrkandena 7 och 8, 2025/26:1774 av Jan Ericson m.fl. (M), 2025/26:1815 av Aylin Nouri och Eva Lindh (båda S), 2025/26:2370 av Malin Östh m.fl. (V) yrkande 11, 2025/26:2832 av Niels Paarup-Petersen m.fl. (C) yrkande 6, 2025/26:3137 av Emma Ahlström Köster (M), 2025/26:3320 av Leila Ali Elmi m.fl. (MP) yrkandena 1-7, 2025/26:3405 av Emma Berginger m.fl. (MP) yrkande 11, 2025/26:3425 av Jan Riise m.fl. (MP) yrkandena 1-3 och 6-8, 2025/26:3623 av Morgan Johansson m.fl. (S) yrkande 69 och 2025/26:3652 av Aylin Nouri m.fl. (S) yrkande 12.
Riksdag decision: The Riksdag said no to approximately 110 proposals in motions on digitalization and postal issues. The Riksdag also tabled the Government's statement dealing with the Swedish National Audit Office's audit report on state strategic digitalization projects for the records, that is, concluded the matter. The motions were submitted during the general motion period 2025 and one of them from 2024. The proposals mainly deal with digital participation, digitalization in public activities, access to electronic communication, artificial intelligence, and postal issues. The Riksdag refers, among other things, to ongoing work and measures taken. In the statement, the Government reports its assessment of the Swedish National Audit Office's findings and recommendations in the audit report State strategic digitalization projects – major shared challenges.
Counter-proposals
- The counter-proposal wants the government to investigate the possibility for guardians to use digital tools for people they have guardianship for. (V) — Reservation 14
Plan vs actual vote
Every party, in chamber order: what its own plan pointed to, how it voted, and the reasoning behind both. Highlighted rows are where the two differ — but a vote that follows the plan is a result too, so the reasoning is there for every party.
The plan requires protection for personal integrity, active work against discrimination and thoughtful discussions about technology's choices because technical shifts shape power relations in society.
Plan documents: party programme
Reasoning and quotes
The plan requires protection for personal integrity, active work against discrimination and thoughtful discussions about technology's choices because technical shifts shape power relations in society. That all public use of AI should be preceded by thorough impact assessments and that public actors receive support to manage the risks follows from that. The plan also wants the tools and platforms of the internet to be built on users' freedom and self-determination.
Original (Swedish)
Planen kräver skydd för den personliga integriteten, aktivt arbete mot diskriminering och genomtänkta diskussioner om teknikens vägval eftersom tekniska skiften formar maktrelationerna i samhället. Att all offentlig AI-användning ska föregås av noggranna konsekvensanalyser och att offentliga aktörer får stöd att hantera riskerna följer av det. Planen vill också att nätets verktyg och plattformar byggs på användarnas frihet och självbestämmande.
- thoughtful technology choicesdecisive»hället i stort behövs det genomtänkta diskussioner om vägvalen.« (party programme ↗)“society as a whole needs thoughtful discussions about the choices.”
- protection for personal integrity»har rätt till skydd av sin personliga integritet.« (party programme ↗)“have the right to protection of their personal integrity.”
- prevent discrimination»Det kräver ett aktivt arbete med att förebygga och« (party programme ↗)“It requires active work to prevent and”
- users' self-determination»ger på användarnas frihet, integritet och självbestämmande, medan« (party programme ↗)“given users' freedom, integrity and self-determination, while”
The party programme demands democratic rules for digital communication and artificial intelligence and clear regulation of AI and algorithms, as the technology otherwise risks damaging trust and democratic dialogue.
Plan documents: party programme
Reasoning and quotes
The party programme demands democratic rules for digital communication and artificial intelligence and clear regulation of AI and algorithms, as the technology otherwise risks damaging trust and democratic dialogue. Requirements for impact assessments before public AI use and support for public actors are the democratic governance the plan seeks. Protection of personal integrity points in the same direction.
Original (Swedish)
Partiprogrammet kräver demokratiska spelregler för digital kommunikation och artificiell intelligens och en tydlig reglering av AI och algoritmer, eftersom tekniken annars riskerar att skada tilliten och det demokratiska samtalet. Krav på konsekvensanalyser före offentlig AI-användning och stöd till offentliga aktörer är den demokratiska styrning planen efterfrågar. Skyddet för den personliga integriteten pekar åt samma håll.
- democratic rules for AIdecisive»Brutna oligopol och demokratiska spelregler för digital kommunikation och artificiell« (party programme ↗)“Broken monopolies and democratic rules for digital communication and artificial”
- clear regulation of AI»Väl fungerande digital inre marknad i EU, med tydlig reglering av artificiell intelligens« (party programme ↗)“Well-functioning digital internal market in the EU, with clear regulation of artificial intelligence”
- protection of integrity»Skydd för den personliga integriteten.« (party programme ↗)“Protection of personal integrity.”
- protection against discrimination»Lika rättigheter och skyldigheter för alla medborgare och starkt skydd mot« (party programme ↗)“Equal rights and duties for all citizens and strong protection against”
The plan takes no explicit position on what the committee actually decided — often because the committee's reason was procedural. The stance is derived and fully cited, but it does not show the party broke a commitment.
The plan states that politics must play an active role in shaping AI and that risks from new powerful technologies must be managed, with respect for fundamental freedoms and rights and integrity.
Plan documents: party programme
Reasoning and quotes
The plan states that politics must play an active role in shaping AI and that risks from new powerful technologies must be managed, with respect for fundamental freedoms and rights and integrity. Requirements for impact analyses before public AI use and support for public actors are precisely such risk management. The plan therefore does not settle for ongoing work being considered sufficient.
Original (Swedish)
Planen slår fast att politiken måste spela en aktiv roll i hur AI formas och att riskerna med nya kraftfulla tekniker måste hanteras, med respekt för grundläggande fri- och rättigheter och integritet. Krav på konsekvensanalyser före offentlig AI-användning och stöd till offentliga aktörer är just en sådan riskhantering. Planen nöjer sig alltså inte med att pågående arbete anses tillräckligt.
- active politics for AIdecisive»Vi i Miljöpartiet anser att politiken måste spela en aktiv roll i hur AI formas och används.« (party programme ↗)“We in the Green Party believe that politics must play an active role in shaping and using AI.”
- risks must be managed»När nya kraftfulla tekniker etableras måste de risker som finns hanteras.« (party programme ↗)“When new powerful technologies become established, the risks that exist must be managed.”
- rights and integrity in AI»utvecklingen av AI ska ske med respekt för grundläggande fri- och rättigheter« (party programme ↗)“the development of AI must occur with respect for fundamental freedoms and rights”
The plan takes no explicit position on what the committee actually decided — often because the committee's reason was procedural. The stance is derived and fully cited, but it does not show the party broke a commitment.
The party programme requires that all legislation be tested against its impact on people's personal integrity and that intrusions be proportionate to the purpose, which in substance is the same requirement as careful impact analyses before public AI use.
Plan documents: party programme · 2022 manifesto
Reasoning and quotes
The party programme requires that all legislation be tested against its impact on people's personal integrity and that intrusions be proportionate to the purpose, which in substance is the same requirement as careful impact analyses before public AI use. The plan also establishes that the constitutional state must exist even in the digital sphere and that discrimination shall be counteracted, which is precisely the risks the counter-proposal points to. That AI is not mentioned by name makes the conclusion derived.
Original (Swedish)
Partiprogrammet kräver att all lagstiftning prövas mot sin påverkan på människors personliga integritet och att intrång står i proportion till syftet, vilket i sak är samma krav som noggranna konsekvensanalyser före offentlig AI-användning. Planen slår också fast att rättssamhället måste finnas även i den digitala sfären och att diskriminering ska motverkas, vilket är precis de risker motförslaget pekar på. Att AI inte nämns vid namn gör slutsatsen härledd.
- integrity review of measuresweak support»All lagstiftning bör prövas mot dess påverkan på människors personliga integritet. Det måste finnas en proportionalitet mellan lagstiftningens inskränkning av medborgarnas integritet och lagarnas syfte.« (party programme ↗)“All legislation should be tested against its impact on people's personal integrity. There must be proportionality between legislation's restriction of citizens' integrity and the laws' purpose.”
- constitutional state even digitallydecisive»Samtidigt måste rättssamhället finnas även i den digitala sfären.« (party programme ↗)“At the same time, the constitutional state must exist even in the digital sphere.”
- integrity as cornerstone»Respekten för den enskildes personliga integritet är en grundsten i varje demokrati.« (party programme ↗)“Respect for the individual's personal integrity is a cornerstone of every democracy.”
- counteract discrimination»Mer måste göras mot diskriminering. Vi vill stärka myndigheternas arbete mot diskriminering och höja ersättningen till dem som utsatts, samt lägga till kön och funktionsnedsättning som hatbrottsgrunder.« (2022 manifesto ↗)“More must be done against discrimination. We want to strengthen the authorities' work against discrimination and raise compensation for those who have been subjected to it, as well as add gender and disability as hate crime grounds.”
flags: citat_svagt
The plan takes no explicit position on what the committee actually decided — often because the committee's reason was procedural. The stance is derived and fully cited, but it does not show the party broke a commitment.
The party programme requires that measures that negatively affect personal integrity are only justified when an opposing interest weighs significantly more heavily and the measure is effective, and that all new bills are analyzed from an integrity perspective.
Plan documents: party programme
Reasoning and quotes
The party programme requires that measures that negatively affect personal integrity are only justified when an opposing interest weighs significantly more heavily and the measure is effective, and that all new bills are analyzed from an integrity perspective. That public AI use shall be preceded by thorough consequence analyses is an application of exactly that review. The plan's purpose limitation for surveillance and its requirement that discrimination shall cost bear other parts of the counter-proposal.
Original (Swedish)
Partiprogrammet kräver att åtgärder som påverkar den personliga integriteten negativt bara är berättigade när ett motstående intresse väger uppenbart tyngre och åtgärden är effektiv, och att alla nya lagförslag analyseras ur integritetssynpunkt. Att offentlig AI-användning ska föregås av noggranna konsekvensanalyser är en tillämpning av just den prövningen. Planens ändamålsbegränsning för övervakning och dess krav på att diskriminering ska kosta bär övriga delar av motförslaget.
- proportionality reviewdecisive»Åtgärder som påverkar den personliga integriteten negativt är endast berättigade om ett motstående intresse väger uppenbart tyngre och åtgärden är effektiv.« (party programme ↗)“Measures that negatively affect personal integrity are only justified if an opposing interest weighs significantly more heavily and the measure is effective.”
- mandatory integrity analysis»Alla nya lagförslag ska analyseras ur integritetssynpunkt.« (party programme ↗)“All new bills shall be analyzed from an integrity perspective.”
- purpose limitation»Övervakning och registrering ska inte användas för andra syften än vad som var de ursprungliga motiven eller vad som var känt för den individ som lämnade uppgifterna.« (party programme ↗)“Surveillance and registration shall not be used for purposes other than what were the original motives or what was known to the individual who provided the information.”
- protection against discrimination»Den som drabbas av diskriminering ska kunna få hjälp att få upprättelse och det ska kosta att diskriminera.« (party programme ↗)“Those affected by discrimination shall be able to get help to get redress and it shall cost to discriminate.”
The plan takes no explicit position on what the committee actually decided — often because the committee's reason was procedural. The stance is derived and fully cited, but it does not show the party broke a commitment.
The plan requires that the public exercise great caution in handling personal data and that personal integrity has strong constitutional protection.
Plan documents: party programme
Reasoning and quotes
The plan requires that the public exercise great caution in handling personal data and that personal integrity has strong constitutional protection. For powerful new technology, the plan says one should proceed extremely carefully and that technology and ethics must not be separated, and before decisions, consequences should be described. Requirements for thorough consequence analyses before public AI use follow from these commitments, as does protection against discrimination.
Original (Swedish)
Planen kräver att det offentliga tillämpar stor försiktighet vid hantering av personuppgifter och att den personliga integriteten har ett starkt grundlagsskydd. För kraftfull ny teknik gäller enligt planen att man ska gå ytterst varsamt fram och att teknik och etik inte får separeras, och inför beslut ska konsekvensbeskrivningar göras. Krav på noggranna konsekvensanalyser före offentlig AI-användning följer av dessa åtaganden, liksom skyddet mot diskriminering.
- strong integrity protectiondecisive»Den personliga integriteten ska ha ett starkt grundlagsskydd« (party programme ↗)“Personal integrity shall have strong constitutional protection”
- technology requires ethical review»Teknik och etik får inte separeras.« (party programme ↗)“Technology and ethics must not be separated.”
- consequences described before decision»Inför planbeslut ska konsekvensbeskrivningar göras« (party programme ↗)“Before planning decisions, consequences should be described”
- protection against discrimination»Grundlagen ska även säkerställa att ingen diskrimineras på grund av kön, ålder« (party programme ↗)“The Constitution shall also ensure that no one is discriminated against on grounds of gender, age”
The plan takes no explicit position on what the committee actually decided — often because the committee's reason was procedural. The stance is derived and fully cited, but it does not show the party broke a commitment.
The plan demands that personal privacy be protected and that all public exercise of power shall be law-governed and never exercised arbitrarily or in a way that discriminates.
Plan documents: party programme · 2022 manifesto
Reasoning and quotes
The plan demands that personal privacy be protected and that all public exercise of power shall be law-governed and never exercised arbitrarily or in a way that discriminates. Demanding consequence analyses before the public puts AI into operation is a way to secure precisely this, and the plan's demand that reforms shall rest on research and proven experience points in the same direction. The plan does not, however, address AI use in administration.
Original (Swedish)
Planen kräver att den personliga integriteten värnas och att all offentlig maktutövning ska vara lagstyrd och aldrig utövas godtyckligt eller på ett sätt som diskriminerar. Att kräva konsekvensanalyser innan det offentliga sätter AI i drift är ett sätt att säkra just detta, och planens krav på att reformer ska vila på forskning och beprövad erfarenhet pekar åt samma håll. Planen behandlar dock inte AI-användning i förvaltningen.
- protect personal privacydecisive»Den personliga integriteten ska värnas, såväl från intrång« (party programme ↗)“Personal privacy shall be protected, both from intrusion”
- no arbitrary exercise of power»utövas godtyckligt, eller på ett sätt som diskriminerar.« (party programme ↗)“exercised arbitrarily, or in a way that discriminates.”
- protection of freedoms and rights»Stärka domstolarnas oberoende och skyddet för medborgarnas fri- och rättigheter« (2022 manifesto ↗)“Strengthen the independence of courts and protection of citizens' freedoms and rights”
flags: citat_blockerat
The plan takes no explicit position on what the committee actually decided — often because the committee's reason was procedural. The stance is derived and fully cited, but it does not show the party broke a commitment.
The plan requires increased transparency about how algorithms affect the spread of content and that this shall not depend on subjective value judgments, and that the state's and large corporations' ability to control information flows according to political interests decreases.
Plan documents: 2022 manifesto
Reasoning and quotes
The plan requires increased transparency about how algorithms affect the spread of content and that this shall not depend on subjective value judgments, and that the state's and large corporations' ability to control information flows according to political interests decreases. It also sees technology's influence over information as a democratic vulnerability. Requiring thorough impact analyses before public AI use aligns with this, even though the plan does not specifically address AI.
Original (Swedish)
Planen kräver ökad transparens kring hur algoritmer påverkar spridningen av innehåll och att detta inte ska bero på subjektiva värderiktningar, samt att statens och storbolagens möjlighet att styra informationsflöden efter politiska intressen minskar. Den ser också teknikens inflytande över informationen som en demokratisk sårbarhet. Krav på noggranna konsekvensanalyser före offentlig AI-användning ligger i linje med detta, även om planen inte behandlar AI särskilt.
- algorithm transparencydecisive»Transparensen kring hur algoritmer påverkar spridningseffekten utifrån innehåll ska öka och ska inte kunna vara avhängigt subjektiva värderiktningar.« (2022 manifesto ↗)“Transparency about how algorithms affect the spread effect based on content shall increase and shall not depend on subjective value judgments.”
- no political control of information flows»Minska möjligheten för staten eller privata storbolag att påverka informationsflöden utifrån politiska intressen« (2022 manifesto ↗)“Reduce the possibility for the state or private large corporations to influence information flows according to political interests”
- democratic vulnerability in technology»Att enskilda aktörer, ofta med bas i utlandet, ges sådant inflytande över vilken information som når fram är ur ett demokratiskt perspektiv en stor sårbarhet.« (2022 manifesto ↗)“That individual actors, often based abroad, are given such influence over what information reaches people is from a democratic perspective a great vulnerability.”
- strengthened position for the individual»Öka användarnas ställning i förhållande till sociala plattformar« (2022 manifesto ↗)“Increase users' position in relation to social platforms”
The plan takes no explicit position on what the committee actually decided — often because the committee's reason was procedural. The stance is derived and fully cited, but it does not show the party broke a commitment.
Share and embed
Sources and documents
- The committee report on riksdagen.se (2025/26:TU8)
- Committee report full text (HD01TU8)
- Per-MP voting data (API)
- Case status (API)
Documents under consideration
- Motion 2024/25:3107 — Internationell samverkan av Morgan Johansson m.fl. (S)
- Motion 2025/26:1082 — Satsning på infrastruktur och digital tillgång i Jönköpings län av Markus Kauppinen m.fl. (S)
- Motion 2025/26:1165 — En säker e-legitimation under statlig kontroll av Adrian Magnusson och Amalia Rud Stenlöf (båda S)
- Motion 2025/26:1323 — Åtgärder för att motverka digitalt utanförskap av Sanna Backeskog m.fl. (S)
- Motion 2025/26:1474 — Skyldighet för mobil- och teleoperatörer att samverka med rättsvårdande myndigheter av Eric Westroth (SD)
- Motion 2025/26:1774 — Risker och möjligheter med artificiell intelligens (AI) av Jan Ericson m.fl. (M)
- Motion 2025/26:1815 — Märkning av AI-genererat material för att motverka desinformation och deepfakes av Aylin Nouri och Eva Lindh (båda S)
- Motion 2025/26:1897 — Spetskompetens och digital utveckling inom tech av Mattias Jonsson och Johan Büser (båda S)
- Motion 2025/26:2297 — Skärpning av lagen om ansvar för elektroniska anslagstavlor - lex Flashback av Markus Selin (S)
- Motion 2025/26:2370 — Fungerande postservice, bredband och telefoni i hela landet av Malin Östh m.fl. (V)
- Motion 2025/26:2391 — Obligatoriskt biometriskt register av Markus Wiechel (SD)
- Motion 2025/26:251 — Förstärkta åtgärder för att bekämpa olagligt innehåll i postförsändelser av Johnny Svedin (SD)
- Motion 2025/26:2591 — Registrering av geodata vid inloggning i bankappar av Mattias Eriksson Falk och Roger Hedlund (båda SD)
- Motion 2025/26:2615 — Ökad transparens på bredbandsmarknaden av Rickard Nordin (C)
- Motion 2025/26:2616 — Enkel rapportering av telefonbedrägerier direkt i telefonen av Rickard Nordin (C)
- Motion 2025/26:2661 — Stärkt krav på mobiloperatörers reservkraft för ett robust samhälle av Cecilia Rönn (L)
- Motion 2025/26:2780 — Transporteffektivt och hållbart samhälle av Nooshi Dadgostar m.fl. (V)
- Motion 2025/26:2832 — Digitalisering för hela Sverige av Niels Paarup-Petersen m.fl. (C)
- Motion 2025/26:2880 — Likvärdig digital infrastruktur i hela Värmland med särskilt fokus på landsbygden av Marléne Lund Kopparklint (M)
- Motion 2025/26:3028 — En fungerande telekommunikation i hela landet av Thomas Ragnarsson (M)
Source: The Swedish Parliament