C_2025_884_2_EN_ACT_part1_v2_OlFHGWKJRcz1XPvgNW2ZVuggMw_112366信息安全资料 .docxVIP

EUROPEANCOMMISSION

Brussels,4.2.2025C(2025)884final

COMMUNICATIONTOTHECOMMISSION

ApprovalofthecontentofthedraftCommunicationfromtheCommission-CommissionGuidelinesonprohibitedartificialintelligencepracticesestablishedbyRegulation(EU)2024/1689(AIAct)

EN EN

COMMUNICATIONTOTHECOMMISSION

ApprovalofthecontentofthedraftCommunicationfromtheCommission-CommissionGuidelinesonprohibitedartificialintelligencepracticesestablishedbyRegulation(EU)2024/1689(AIAct)

Regulation(EU)2024/1689oftheEuropeanParliamentandtheCouncilof13June2024layingdownharmonisedrulesonartificialintelligenceandamendingcertainregulations(‘theAIAct’)1enteredintoforceon1August2024.TheAIActlaysdownharmonisedrulesfortheplacingonthemarket,puttingintoservice,anduseofartificialintelligence(‘AI’)intheUnion.ItsaimistopromoteinnovationinandtheuptakeofAI,whileensuringahighlevelofprotectionofhealth,safetyandfundamentalrightsintheUnion,includingdemocracyandtheruleoflaw.

TheAIActfollowsarisk-basedapproach,classifyingAIsystemsintodifferentriskcategories,oneofwhichconsistsofAIpracticesposingunacceptableriskstofundamentalrightsandUnionvalueswhichareprohibitedunderArticle5AIAct.

PursuanttoArticle96(1)(b)AIAct,theCommissionistoadoptguidelinesonthepracticalimplementationofthepracticesprohibitedunderArticle5AIAct.TheDraftGuidelinesannexedtothisCommunicationaimtoincreaselegalclarityandtoprovideinsightsintotheCommission’sinterpretationoftheprohibitionsinArticle5AIActwithaviewtoensuringtheirconsistent,effectiveanduniformapplication.

SuchclarityisessentialforprovidersanddeployersofAIsystems,sincetheprohibitionsaredirectlyapplicableasfrom2February2025.Inviewofthisimminententryintoapplicationandthetimenecessaryforthetranslationoftheguidelinesintoalllanguag