{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,11]],"date-time":"2026-10-11T16:47:09Z","timestamp":1791737229611,"version":"4.3.4"},"reference-count":0,"publisher":"Association for the Advancement of Artificial Intelligence (AAAI)","issue":"4","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["AIES"],"abstract":"<jats:p>The EU AI Act will begin full enforcement for high-risk\napplications in August 2026. For emerging economies, this\npresents a tough choice: adopt a regulatory framework\nmainly aimed at large firms with strong compliance\nresources, or delay action, leaving citizens vulnerable to\nunregulated algorithmic decision-making. Our team decided\nto explore whether a third option might exist. Over ten\nmonths, researchers from computer engineering, software\nengineering, law, medicine, and international trade\nexamined how AI governance works in seven sectors: Fintech,\nhealthcare, education, energy, media, logistics, and AI\nliteracy and compared their findings with six national\nregulatory approaches: the EU, US, UK, Singapore, South\nKorea, and Japan.\nWe found a structural mismatch in how risk categories are\napplied. The EU AI Act places the oversight of an\nautonomous nuclear plant on the same level as predicting\nenergy production in a wind farm, despite the very\ndifferent consequences of failure. This means that a small\nenergy startup working on weather prediction models faces\nthe same compliance requirements as a vendor of nuclear\nsafety systems. In the six jurisdictions we reviewed, we\ndid not find a way to adjust regulatory obligations based\non the actual innovation value of a technology or its\ndomestic origin.\nThe US model seemed more promising at first because it\noperates on a sector-by-sector basis. However, its\ndependence on voluntary commitments allows companies to\nmostly avoid compliance, which defeats the purpose for\neconomies wanting enforceable baseline protections.\nWe took a different route by creating a weighted risk\nscoring system. In this model, innovation potential and\ndomestic development capacity serve as negative weight\nparameters, lessening the regulatory burden for\napplications that provide clear value. At the same time, we\nestablished a strict safety floor. Any system that scores\nat maximum impact severity will receive the most stringent\noversight, regardless of other factors. The guiding\nprinciple was clear: no matter how innovative an AI system\nis, if it controls a nuclear reactor, it must be regulated\naccordingly.\nFor the financial sector, we introduced what we call a \"Too\nbig to fail AI\" parameter. This accounts for the risk that\na single algorithmic system's failure can have a ripple\neffect through interconnected institutions. To our\nknowledge, no current framework tries to measure this\nsystemic aspect.\nInstead of using a single sandbox model across all sectors,\nwe created sector-specific mechanisms. Healthcare follows a\nthree-stage process: simulation, controlled clinical\ndeployment, and limited market release. Energy systems must\ncomplete 10,000 hours of digital twin validation before\nlinking to the live grid. Fintech employs a graduated\nscaling approach from micro-pilots to nationwide rollout.\nEach design came from the risk characteristics specific to\nits sector.\nOne unexpected finding was the extent to which policymakers\nbelieved the EU framework could be adopted with minimal\nadjustments. Few assessed whether their domestic market\nstructures, often led by SMEs with limited compliance\ncapacity, could realistically manage the costs involved.\nOur graduated obligation model was designed specifically to\ntackle this issue, differentiating requirements across\nenterprise tiers based on organizational capacity.<\/jats:p>","DOI":"10.1609\/aies.v9i4.43331","type":"journal-article","created":{"date-parts":[[2026,10,11]],"date-time":"2026-10-11T16:33:59Z","timestamp":1791736439000},"page":"2574-2583","source":"Crossref","is-referenced-by-count":0,"title":["Beyond Copy-Paste Regulation: Innovation-Weighted Risk Classification for AI Governance in Emerging Economies"],"prefix":"10.1609","volume":"9","author":[{"given":"Egemen Bora","family":"Tun\u00e7arslan","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ece Melisa","family":"\u015eerik","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Samed","family":"\u00c7ay\u0131r","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"9382","published-online":{"date-parts":[[2026,10,11]]},"container-title":["Proceedings of the AAAI\/ACM Conference on AI, Ethics, and Society"],"original-title":[],"link":[{"URL":"https:\/\/ojs.aaai.org\/index.php\/AIES\/article\/download\/43331\/50891","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/ojs.aaai.org\/index.php\/AIES\/article\/download\/43331\/50891","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,11]],"date-time":"2026-10-11T16:33:59Z","timestamp":1791736439000},"score":1,"resource":{"primary":{"URL":"https:\/\/ojs.aaai.org\/index.php\/AIES\/article\/view\/43331"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,10,11]]},"references-count":0,"journal-issue":{"issue":"4","published-online":{"date-parts":[[2026,10,11]]}},"URL":"https:\/\/doi.org\/10.1609\/aies.v9i4.43331","relation":{},"ISSN":["3065-8365"],"issn-type":[{"value":"3065-8365","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10,11]]}}}