{"id":77926,"date":"2026-10-02T11:34:52","date_gmt":"2026-10-02T09:34:52","guid":{"rendered":"https:\/\/addcomply.com\/?p=77926"},"modified":"2026-10-02T12:30:07","modified_gmt":"2026-10-02T10:30:07","slug":"high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act","status":"publish","type":"post","link":"https:\/\/addcomply.com\/sv\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/","title":{"rendered":"H\u00f6grisk eller inte? S\u00e5 klassificerar ni era AI-system enligt AI-f\u00f6rordningen"},"content":{"rendered":"<h2 id=\"why-classification-is-the-first-step\" data-lfile-container=\"829de1bf-4bbf-473f-bb5e-4067169dd48e\" data-pm-slice=\"1 1 []\">Varf\u00f6r klassificering \u00e4r f\u00f6rsta steget<\/h2>\n<p data-lfile-container=\"829de1bf-4bbf-473f-bb5e-4067169dd48e\" data-pm-slice=\"1 1 []\">AI-f\u00f6rordningen bygger p\u00e5 en riskbaserad modell: ju h\u00f6gre risk ett AI-system inneb\u00e4r f\u00f6r m\u00e4nniskors h\u00e4lsa, s\u00e4kerhet och grundl\u00e4ggande r\u00e4ttigheter, desto str\u00e4ngare krav st\u00e4lls p\u00e5 dig som leverant\u00f6r (Provider) och anv\u00e4ndare (Deployer). Att klassificera AI-system r\u00e4tt handlar d\u00e4rf\u00f6r om att f\u00f6rst ta reda p\u00e5 vilken roll och vilken riskniv\u00e5 varje system har \u2013 det \u00e4r den processen vi kallar klassificering.<\/p>\n<p>Att klassificera dina AI-system r\u00e4tt \u00e4r viktigt b\u00e5de f\u00f6r att veta vilka \u00e5tg\u00e4rder som kr\u00e4vs f\u00f6r att uppfylla lagstiftningens krav och f\u00f6r att undvika on\u00f6diga kostnader. Ett system som felaktigt bed\u00f6ms som l\u00e5grisk riskerar sanktionsavgifter om tillsynsmyndigheten senare kommer fram till att det borde ha klassats som h\u00f6grisk. Ett system som felaktigt klassificeras som h\u00f6grisk belastas i st\u00e4llet med on\u00f6diga krav p\u00e5 riskhantering, teknisk dokumentation och bed\u00f6mning av \u00f6verensst\u00e4mmelse med lagstiftningen.<\/p>\n<p>Den h\u00e4r artikeln g\u00e5r igenom AI-f\u00f6rordningens fyra riskniv\u00e5er, vad som avg\u00f6r om ett system \u00e4r h\u00f6grisk, och hur du kommer ig\u00e5ng med att klassificera AI-systemen i din organisation.<\/p>\n<h2 id=\"the-four-risk-levels\">De fyra riskniv\u00e5erna<\/h2>\n<p>AI-f\u00f6rordningen delar in AI-system i fyra niv\u00e5er:<\/p>\n<div class=\"ac-table-wrap\">\n<table>\n<thead>\n<tr>\n<th>Riskniv\u00e5<\/th>\n<th>Vad det inneb\u00e4r<\/th>\n<th>Exempel<\/th>\n<th>Krav<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Oacceptabel risk<\/td>\n<td>F\u00f6rbjuden anv\u00e4ndning<\/td>\n<td>Social po\u00e4ngs\u00e4ttning, manipulativ AI, biometrisk mass\u00f6vervakning i realtid<\/td>\n<td>F\u00f6rbjudet sedan 2 februari 2025.<\/td>\n<\/tr>\n<tr>\n<td>H\u00f6g risk<\/td>\n<td>Full efterlevnad kr\u00e4vs innan systemet f\u00e5r s\u00e4ttas i drift<\/td>\n<td>Rekrytering, kreditbed\u00f6mning, biometrisk identifiering, utbildning, kritisk infrastruktur<\/td>\n<td>Riskhantering, teknisk dokumentation, m\u00e4nsklig \u00f6versyn, konformitetsbed\u00f6mning g\u00e4ller fr\u00e5n 2 december 2027.<\/td>\n<\/tr>\n<tr>\n<td>Begr\u00e4nsad risk<\/td>\n<td>Transparens mot anv\u00e4ndaren<\/td>\n<td>Chattbottar, AI-genererat inneh\u00e5ll (deepfakes)<\/td>\n<td>Information om att man interagerar med eller exponeras f\u00f6r AI g\u00e4ller fr\u00e5n 2 augusti och 2 december 2026.<\/td>\n<\/tr>\n<tr>\n<td>Minimal risk<\/td>\n<td>Ingen s\u00e4rskild reglering<\/td>\n<td>Spamfilter, AI i dataspel<\/td>\n<td>Frivilliga uppf\u00f6randekoder rekommenderas.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>M\u00e5nga AI-system p\u00e5 marknaden hamnar i kategorierna minimal eller begr\u00e4nsad risk och h\u00e4r finns vissa krav som beh\u00f6ver g\u00f6ras eller andra frivilliga uppf\u00f6randekoder som endast \u00e4r en rekommendation. Det \u00e4r h\u00f6g risk-kategorin eller n\u00e4ra h\u00f6g-risk som kr\u00e4ver mest arbete och d\u00e4r gr\u00e4nsdragningen \u00e4r sv\u00e5rast att g\u00f6ra.<\/p>\n<h2 id=\"unacceptable-risk-prohibited-ai-use\">Oacceptabel risk \u2013 f\u00f6rbjuden AI anv\u00e4ndning<\/h2>\n<p>Sedan den 2 februari 2025 \u00e4r sex typer av AI-anv\u00e4ndning f\u00f6rbjudna inom EU:<\/p>\n<ul>\n<li>Manipulativ eller vilseledande AI som m\u00e5lmedvetet p\u00e5verkar m\u00e4nniskors beteende eller k\u00e4nslor f\u00f6r att orsaka skada<\/li>\n<li>Exploatering av utsatta grupper kopplat till \u00e5lder, funktionsneds\u00e4ttning eller socioekonomisk situation<\/li>\n<li>Social po\u00e4ngs\u00e4ttning \u2013 rangordning av personer baserat p\u00e5 beteende eller socioekonomisk status<\/li>\n<li>AI som f\u00f6rutser risk f\u00f6r att beg\u00e5 brott genom predikativ polisi\u00e4r profilering baserad enbart p\u00e5 personliga drag eller datam\u00f6nster<\/li>\n<li>K\u00e4nsloigenk\u00e4nning p\u00e5 arbetsplatsen eller i skolan<\/li>\n<li>Biometrisk identifiering i realtid p\u00e5 allm\u00e4n plats, med undantag f\u00f6r brottsbek\u00e4mpning under strikta villkor<\/li>\n<\/ul>\n<p>Tack vare den digitala omnibusen som r\u00f6stades igenom i juli 2026 tillkommer ytterligare tv\u00e5 kategorier av f\u00f6rbjuden AI-anv\u00e4ndning den 2 december 2026:<\/p>\n<ul>\n<li><span class=\"a_GcMg font-feature-liga-off font-feature-clig-off font-feature-calt-off text-decoration-none text-strikethrough-none\">Nudify-appar: Det \u00e4r f\u00f6rbjudet med AI-appar som skapar o\u00f6nskade och icke-samtyckta avkl\u00e4dda bilder (intim deepfake-material).<\/span><span class=\"a_GcMg font-feature-liga-off font-feature-clig-off font-feature-calt-off text-decoration-none text-strikethrough-none\"> AI apps that create unwanted and non-consensual undressed images (intimate deepfake material) are prohibited.<\/span><\/li>\n<li><span class=\"a_GcMg font-feature-liga-off font-feature-clig-off font-feature-calt-off text-decoration-none text-strikethrough-none\">Child sexual abuse:<\/span><span class=\"a_GcMg font-feature-liga-off font-feature-clig-off font-feature-calt-off text-decoration-none text-strikethrough-none\"> Sexuella \u00f6vergrepp mot barn: Sk\u00e4rpta och ut\u00f6kade f\u00f6rbud mot AI-genererat material som r\u00f6r sexuella \u00f6vergrepp mot barn (CSAM)<\/span><\/li>\n<\/ul>\n<p>Om din organisation upphandlar eller bygger AI-l\u00f6sningar: screena leverant\u00f6rer mot listan innan avtal tecknas. Ett system som byggts f\u00f6r ett annat syfte kan \u200b\u00e4nd\u00e5 hamna h\u00e4r om det faktiskt kan anv\u00e4nds p\u00e5 ett s\u00e4tt som faller inom n\u00e5got av ovan angivna omr\u00e5den. De \u00e4r b\u00e5de anv\u00e4ndningen och den ursprungliga avsikten som avg\u00f6r.<\/p>\n<p>I den h\u00e4r artikeln kan du l\u00e4sa mer om <a href=\"https:\/\/addcomply.com\/sv\/prohibited-ai-systems-eu-ai-act\/\">f\u00f6rbjuden AI-anv\u00e4ndning.<\/a>.<\/p>\n<h2 id=\"high-risk\">H\u00f6g risk<\/h2>\n<p>Ett AI-system klassas som h\u00f6grisk p\u00e5 ett av tv\u00e5 s\u00e4tt:<\/p>\n<div class=\"ac-table-wrap\">\n<table>\n<thead>\n<tr>\n<th>V\u00e4g<\/th>\n<th>R\u00e4ttslig grund<\/th>\n<th>Vad som avses<\/th>\n<th>Ny deadline<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Produkt<\/td>\n<td>Article 6(1) + Annex I<\/td>\n<td>AI som ing\u00e5r i eller \u00e4r en s\u00e4kerhetskomponent i en produkt som redan omfattas av EU:s harmoniseringslagstiftning (t.ex. medicinteknik, fordon, leksaker) och kr\u00e4ver tredjepartsbed\u00f6mning.<\/td>\n<td>2 augusti 2028<\/td>\n<\/tr>\n<tr>\n<td>Anv\u00e4ndningsfall<\/td>\n<td>Artikel 6(2) + bilaga III<\/td>\n<td>Frist\u00e5ende AI-system som anv\u00e4nds i s\u00e4rskilt utpekade h\u00f6griskomr\u00e5den: biometri, kritisk infrastruktur, utbildning, anst\u00e4llning, v\u00e4sentliga tj\u00e4nster, brottsbek\u00e4mpning, migration, r\u00e4ttsskipning<\/td>\n<td>2 december 2027<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>B\u00e5da deadlines sk\u00f6ts fram genom EU:s \u201cDigital Omnibus\u201d i 12 respektive 16 m\u00e5nader senare \u00e4n ursprungsplanen. Kraven \u00e4r i stort sett of\u00f6r\u00e4ndrade, det \u00e4r bara tidpunkten f\u00f6r n\u00e4r de m\u00e5ste vara uppfyllda som flyttats fram.<\/p>\n<h3 id=\"what-makes-a-system-a-safety-component\">Vad g\u00f6r ett system till en s\u00e4kerhetskomponent?<\/h3>\n<p>EU-kommissionen publicerade i maj 2026 <a href=\"https:\/\/digital-strategy.ec.europa.eu\/en\/library\/draft-commission-guidelines-classification-high-risk-ai-systems\" target=\"_blank\" rel=\"noopener noreferrer nofollow\">utkast till riktlinjer f\u00f6r klassificering av h\u00f6grisksystem<\/a> som ett hj\u00e4lpmedel f\u00f6r att g\u00f6ra bed\u00f6mningen mer f\u00f6ruts\u00e4gbar och enhetlig i alla 27 medlemsl\u00e4nder. En s\u00e4kerhetskomponent \u00e4r ett system som har n\u00e5gon av f\u00f6ljande funktionalitet:<\/p>\n<ul>\n<li>Felfunktion: ett fel eller haveri i systemet skulle kunna \u00e4ventyra h\u00e4lsa, s\u00e4kerhet eller egendom, \u00e4ven om det inte var det avsedda syftet (t.ex. AI som styr en hissd\u00f6rrs timing)<\/li>\n<li>S\u00e4kerhetsfunktion: systemets uttalade syfte \u00e4r att f\u00f6rebygga eller mildra risker f\u00f6r h\u00e4lsa, s\u00e4kerhet eller egendom (t.ex. kollisionsdetektering i fordon)<\/li>\n<\/ul>\n<p>S\u00e5, AI-system som med funktioner f\u00f6r prestandaoptimering, komfort och effektivitet r\u00e4knas inte som s\u00e4kerhetsfunktioner och \u00e4r d\u00e4rmed inte att betrakta som ett h\u00f6g-risk system enligt AI-f\u00f6rordningen., s\u00e5vida inte ett fel i dem skapar en s\u00e4kerhetsrisk.<\/p>\n<h3 id=\"intended-purpose-is-decisive-document-it-carefully\">Avsett syfte avg\u00f6r \u2013 dokumentera det noga<\/h3>\n<p>Klassificeringen styrs av vad leverant\u00f6ren anger att systemet \u00e4r avsett f\u00f6r eller kan anv\u00e4ndas till. Det som r\u00e4knas in: bruksanvisningar, teknisk dokumentation och marknadsf\u00f6ringsmaterial. Ett generellt AI-system utan tydliga begr\u00e4nsningar riskerar att bed\u00f6mas utifr\u00e5n de anv\u00e4ndningsfall det skulle kunna anv\u00e4ndas f\u00f6r, dvs \u00e4ven inom h\u00f6g risk. Tredje parter som \u00e4ndrar det avsedda syftet \u00e4rver leverant\u00f6rens skyldigheter.<\/p>\n<p>Kan du argumentera f\u00f6r att ditt system, trots att det anv\u00e4nds inom ett h\u00f6grisk omr\u00e5de, inte utg\u00f6r risk f\u00f6r m\u00e4nniskors h\u00e4lsa, s\u00e4kerhet och grundl\u00e4ggande r\u00e4ttigheter? D\u00e5 kan du g\u00f6ra g\u00e4llande att du har m\u00f6jlighet att undantas fr\u00e5n kraven f\u00f6r h\u00f6grisksystem.<\/p>\n<h2 id=\"limited-risk-transparency-requirements\">Begr\u00e4nsad risk \u2013 transparenskrav<\/h2>\n<p>System som interagerar direkt med m\u00e4nniskor eller genererar inneh\u00e5ll omfattas av transparenskrav enligt artikel 50 i AI-f\u00f6rordningen och som g\u00e4ller fr\u00e5n den 2 augusti 2026. Transparenskravet omfattar f\u00f6ljande omr\u00e5den:<\/p>\n<ul>\n<li>Chattbottar och AI-assistenter m\u00e5ste tala om att anv\u00e4ndaren interagerar med AI<\/li>\n<li>AI-genererat eller AI-manipulerat inneh\u00e5ll s\u00e5som bild, ljud, video (\u201cdeepfakes\u201d) \u2013 ska m\u00e4rkas som s\u00e5dant<\/li>\n<li>System f\u00f6r k\u00e4nsloigenk\u00e4nning eller biometrisk kategorisering m\u00e5ste informera de personer som exponeras<\/li>\n<li>AI-genererade texter som publiceras f\u00f6r att informera allm\u00e4nheten om fr\u00e5gor av allm\u00e4nt intresse ska m\u00e4rkas \u2013 om inte inneh\u00e5llet genomg\u00e5tt m\u00e4nsklig redaktionell granskning<\/li>\n<\/ul>\n<p>Transparenskraven g\u00e4ller oavsett hur systemet i \u00f6vrigt klassificeras, det \u00e4r sj\u00e4lva interaktionen med m\u00e4nniskor som utl\u00f6ser kravet, inte systemets \u00f6vergripande riskniv\u00e5.<\/p>\n<h2 id=\"how-to-classify-your-ai-systems-step-by-step\">S\u00e5 klassificerar ni era AI-system \u2013 steg f\u00f6r steg<\/h2>\n<ol>\n<li>Inventera. Lista varje AI-system som anv\u00e4nds i organisationen \u2013 ink\u00f6pt, egenbyggt eller inbyggt i annan programvara \u2013 on-prem, cloud, skugg-AI.<\/li>\n<li>Best\u00e4m er roll f\u00f6r varje system. \u00c4r ni leverant\u00f6r (Provider) som utvecklar eller s\u00e4tter systemet p\u00e5 marknaden i eget namn eller \u00e4r ni tillhandah\u00e5llare (Deployer) som anv\u00e4nder ett system n\u00e5gon annan utvecklat. Skyldigheterna skiljer sig \u00e5t.<\/li>\n<li>Screena mot f\u00f6rbudslistan. Uteslut f\u00f6rst att systemet faller inom n\u00e5gon av de f\u00f6rbjudna praktikerna.<\/li>\n<li>Testa mot artikel 6 om ni har ett h\u00f6gris-system. Ing\u00e5r systemet i en produkt eller anv\u00e4nds det inom ett av h\u00f6griskomr\u00e5dena i bilaga III?<\/li>\n<li>Granska det avsedda syftet. G\u00e5 igenom bruksanvisningar, teknisk dokumentation. St\u00e4mmer ert anv\u00e4ndningsomr\u00e5de med det systemet \u00e4r till f\u00f6r?<\/li>\n<li>Dokumentera bed\u00f6mningen. \u00c4ven system som landar i minimal eller begr\u00e4nsad risk b\u00f6r ha en kort dokumentation sparad. Tillsynsmyndigheter kan komma att fr\u00e5ga efter den.<\/li>\n<li>S\u00e4tt datum. Notera vilken deadline som g\u00e4ller f\u00f6r just ert system och koppla den till er interna compliance-plan.<\/li>\n<\/ol>\n<h2 id=\"summary-and-next-steps\">Sammanfattning och n\u00e4sta steg<\/h2>\n<p>De flesta organisationer anv\u00e4nder redan fler AI-system \u00e4n de har koll p\u00e5 och det \u00e4r klassificeringen som avg\u00f6r vilka compliancekrav som v\u00e4ntar f\u00f6r respektive system. S\u00e5 b\u00f6rja i god tid med inventeringen och klassificeringen. Den l\u00e4gger grunden f\u00f6r resten av arbetet och g\u00f6r det m\u00f6jligt att prioritera r\u00e4tt system f\u00f6rst! Det h\u00e4r \u00e4r ett omfattande arbete och mycket detaljer att h\u00e5lla ordning p\u00e5. I v\u00e5r plattform AI Governance Starter f\u00e5r du st\u00f6d f\u00f6r b\u00e5de inventering, klassning, riskbed\u00f6mning och \u00e5tg\u00e4rdsplan. Boka g\u00e4rna en demo s\u00e5 ber\u00e4ttar vi mer!<\/p>","protected":false},"excerpt":{"rendered":"<p>Is your AI system high-risk, low-risk or prohibited? Here&#8217;s how to classify AI systems under the AI Act, with concrete steps and the deadlines you need to keep in mind.<\/p>","protected":false},"author":2,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_et_pb_use_builder":"","_et_pb_old_content":"","_et_gb_content_width":"","footnotes":"","_links_to":"","_links_to_target":""},"categories":[8],"tags":[],"class_list":["post-77926","post","type-post","status-publish","format-standard","hentry","category-insights"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v28.7 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>High-risk or not? How to classify your AI systems under the AI Act &#8212; AddComply<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/addcomply.com\/sv\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/\" \/>\n<meta property=\"og:locale\" content=\"sv_SE\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"High-risk or not? How to classify your AI systems under the AI Act &#8212; AddComply\" \/>\n<meta property=\"og:description\" content=\"Is your AI system high-risk, low-risk or prohibited? Here&#039;s how to classify AI systems under the AI Act, with concrete steps and the deadlines you need to keep in mind.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/addcomply.com\/sv\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/\" \/>\n<meta property=\"og:site_name\" content=\"AddComply\" \/>\n<meta property=\"article:published_time\" content=\"2026-10-02T09:34:52+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-10-02T10:30:07+00:00\" \/>\n<meta name=\"author\" content=\"Emeli Olaison\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Skriven av\" \/>\n\t<meta name=\"twitter:data1\" content=\"Emeli Olaison\" \/>\n\t<meta name=\"twitter:label2\" content=\"Ber\u00e4knad l\u00e4stid\" \/>\n\t<meta name=\"twitter:data2\" content=\"7 minuter\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/addcomply.com\\\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/addcomply.com\\\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\\\/\"},\"author\":{\"name\":\"Emeli Olaison\",\"@id\":\"https:\\\/\\\/addcomply.com\\\/#\\\/schema\\\/person\\\/c0661304e86a75925f9e4f83b821ba0f\"},\"headline\":\"High-risk or not? How to classify your AI systems under the AI Act\",\"datePublished\":\"2026-10-02T09:34:52+00:00\",\"dateModified\":\"2026-10-02T10:30:07+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/addcomply.com\\\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\\\/\"},\"wordCount\":1389,\"publisher\":{\"@id\":\"https:\\\/\\\/addcomply.com\\\/#organization\"},\"articleSection\":[\"News &amp; Insights\"],\"inLanguage\":\"sv-SE\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/addcomply.com\\\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\\\/\",\"url\":\"https:\\\/\\\/addcomply.com\\\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\\\/\",\"name\":\"High-risk or not? How to classify your AI systems under the AI Act &#8212; AddComply\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/addcomply.com\\\/#website\"},\"datePublished\":\"2026-10-02T09:34:52+00:00\",\"dateModified\":\"2026-10-02T10:30:07+00:00\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/addcomply.com\\\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\\\/#breadcrumb\"},\"inLanguage\":\"sv-SE\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/addcomply.com\\\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\\\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/addcomply.com\\\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/addcomply.com\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"High-risk or not? How to classify your AI systems under the AI Act\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/addcomply.com\\\/#website\",\"url\":\"https:\\\/\\\/addcomply.com\\\/\",\"name\":\"AddComply - Solve your EU AI Act challenges\",\"description\":\"The only AI Governance Agent you need\",\"publisher\":{\"@id\":\"https:\\\/\\\/addcomply.com\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/addcomply.com\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"sv-SE\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/addcomply.com\\\/#organization\",\"name\":\"AddComply\",\"url\":\"https:\\\/\\\/addcomply.com\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"sv-SE\",\"@id\":\"https:\\\/\\\/addcomply.com\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/addcomply.com\\\/wp-content\\\/uploads\\\/2025\\\/02\\\/AddComply-logo-black.svg\",\"contentUrl\":\"https:\\\/\\\/addcomply.com\\\/wp-content\\\/uploads\\\/2025\\\/02\\\/AddComply-logo-black.svg\",\"width\":1,\"height\":1,\"caption\":\"AddComply\"},\"image\":{\"@id\":\"https:\\\/\\\/addcomply.com\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/www.linkedin.com\\\/company\\\/addcomply\\\/\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/addcomply.com\\\/#\\\/schema\\\/person\\\/c0661304e86a75925f9e4f83b821ba0f\",\"name\":\"Emeli Olaison\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"High-risk or not? How to classify your AI systems under the AI Act &#8212; AddComply","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/addcomply.com\/sv\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/","og_locale":"sv_SE","og_type":"article","og_title":"High-risk or not? How to classify your AI systems under the AI Act &#8212; AddComply","og_description":"Is your AI system high-risk, low-risk or prohibited? Here's how to classify AI systems under the AI Act, with concrete steps and the deadlines you need to keep in mind.","og_url":"https:\/\/addcomply.com\/sv\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/","og_site_name":"AddComply","article_published_time":"2026-10-02T09:34:52+00:00","article_modified_time":"2026-10-02T10:30:07+00:00","author":"Emeli Olaison","twitter_card":"summary_large_image","twitter_misc":{"Skriven av":"Emeli Olaison","Ber\u00e4knad l\u00e4stid":"7 minuter"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/addcomply.com\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/#article","isPartOf":{"@id":"https:\/\/addcomply.com\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/"},"author":{"name":"Emeli Olaison","@id":"https:\/\/addcomply.com\/#\/schema\/person\/c0661304e86a75925f9e4f83b821ba0f"},"headline":"High-risk or not? How to classify your AI systems under the AI Act","datePublished":"2026-10-02T09:34:52+00:00","dateModified":"2026-10-02T10:30:07+00:00","mainEntityOfPage":{"@id":"https:\/\/addcomply.com\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/"},"wordCount":1389,"publisher":{"@id":"https:\/\/addcomply.com\/#organization"},"articleSection":["News &amp; Insights"],"inLanguage":"sv-SE"},{"@type":"WebPage","@id":"https:\/\/addcomply.com\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/","url":"https:\/\/addcomply.com\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/","name":"High-risk or not? How to classify your AI systems under the AI Act &#8212; AddComply","isPartOf":{"@id":"https:\/\/addcomply.com\/#website"},"datePublished":"2026-10-02T09:34:52+00:00","dateModified":"2026-10-02T10:30:07+00:00","breadcrumb":{"@id":"https:\/\/addcomply.com\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/#breadcrumb"},"inLanguage":"sv-SE","potentialAction":[{"@type":"ReadAction","target":["https:\/\/addcomply.com\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/addcomply.com\/high-risk-or-not-how-to-classify-your-ai-systems-under-the-ai-act\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/addcomply.com\/"},{"@type":"ListItem","position":2,"name":"High-risk or not? How to classify your AI systems under the AI Act"}]},{"@type":"WebSite","@id":"https:\/\/addcomply.com\/#website","url":"https:\/\/addcomply.com\/","name":"AddComply - Solve your EU AI Act challenges","description":"The only AI Governance Agent you need","publisher":{"@id":"https:\/\/addcomply.com\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/addcomply.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"sv-SE"},{"@type":"Organization","@id":"https:\/\/addcomply.com\/#organization","name":"AddComply","url":"https:\/\/addcomply.com\/","logo":{"@type":"ImageObject","inLanguage":"sv-SE","@id":"https:\/\/addcomply.com\/#\/schema\/logo\/image\/","url":"https:\/\/addcomply.com\/wp-content\/uploads\/2025\/02\/AddComply-logo-black.svg","contentUrl":"https:\/\/addcomply.com\/wp-content\/uploads\/2025\/02\/AddComply-logo-black.svg","width":1,"height":1,"caption":"AddComply"},"image":{"@id":"https:\/\/addcomply.com\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.linkedin.com\/company\/addcomply\/"]},{"@type":"Person","@id":"https:\/\/addcomply.com\/#\/schema\/person\/c0661304e86a75925f9e4f83b821ba0f","name":"Emeli Olaison"}]}},"_links":{"self":[{"href":"https:\/\/addcomply.com\/sv\/wp-json\/wp\/v2\/posts\/77926","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/addcomply.com\/sv\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/addcomply.com\/sv\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/addcomply.com\/sv\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/addcomply.com\/sv\/wp-json\/wp\/v2\/comments?post=77926"}],"version-history":[{"count":6,"href":"https:\/\/addcomply.com\/sv\/wp-json\/wp\/v2\/posts\/77926\/revisions"}],"predecessor-version":[{"id":77932,"href":"https:\/\/addcomply.com\/sv\/wp-json\/wp\/v2\/posts\/77926\/revisions\/77932"}],"wp:attachment":[{"href":"https:\/\/addcomply.com\/sv\/wp-json\/wp\/v2\/media?parent=77926"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/addcomply.com\/sv\/wp-json\/wp\/v2\/categories?post=77926"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/addcomply.com\/sv\/wp-json\/wp\/v2\/tags?post=77926"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}