Trend AnalysisInterdisciplinary

Data Ethics and Responsible AI Governance Frameworks

As AI systems make decisions about loan approvals, hiring, medical diagnoses, and criminal sentencing, the question shifts from 'Can AI do this?' to 'Should AI do this, and under what governance conditions?' The 2024-2025 literature reveals a global governance landscape that is fragmented, rapidly evolving, and struggling to keep pace with deployment.

By Sean K.S. Shin
This blog summarizes research trends based on published paper abstracts. Specific numbers or findings may contain inaccuracies. For scholarly rigor, always consult the original papers cited in each post.

Why It Matters

AI governance is no longer a theoretical exercise. The EU AI Act entered into force in 2024, creating the world's first comprehensive legal framework for AI regulation. China's AI governance regulations have been operational since 2023. The US executive order on AI safety (October 2023) established federal requirements for AI risk assessment. Brazil, India, Singapore, and dozens of other nations are developing their own approaches.

What makes AI governance uniquely challenging is the technology's dual nature. The same facial recognition system that helps find missing children enables mass surveillance. The same language model that assists students with learning enables industrial-scale disinformation. The same medical AI that improves diagnostic accuracy in well-resourced hospitals may perform poorly on populations underrepresented in training data, widening health disparities.

Data ethicsโ€”the principles governing how data is collected, used, shared, and protectedโ€”underlies all AI governance. AI systems are built on data, and the ethical quality of the data determines the ethical quality of the system. Training data that reflects historical discrimination produces discriminatory AI. Data collected without informed consent violates autonomy regardless of how the AI uses it. Data governance and AI governance are inseparable.

The Science

EU Compliance Framework for Universities

Britchenko and Lysiak (2025) develop a compliance-capability framework for universities navigating EU data governance and AI ethics regulations. Universities occupy a unique position: they are simultaneously data holders (student records, research data), data users (analytics, AI-assisted administration), and data intermediaries (research data sharing)โ€”making compliance particularly complex.

The framework maps the intersection of the General Data Protection Regulation (GDPR), the Data Governance Act, the AI Act, and sector-specific education regulations. Key finding: most universities lack the organizational capability to comply with the full stack of EU data governance requirements. Compliance requires not just legal knowledge but data infrastructure, staff training, ethical review processes, and governance structures that most institutions have not yet built.

The practical implication extends beyond universities: any organization operating as a complex data ecosystem (hospitals, government agencies, multinational corporations) faces the same compliance-capability gap.

Malaysia's AI Governance Implementation

Mat Taib et al. (2025) document lessons from Malaysia's implementation of its responsible AI governance frameworkโ€”providing a rare empirical study of what happens when AI governance moves from policy document to operational reality.

Malaysia's framework emphasizes seven principles: fairness, reliability, privacy, inclusiveness, transparency, accountability, and human oversight. The study evaluates implementation across government agencies and private sector organizations, finding significant variation in adoption. Large technology companies implement governance mechanisms relatively quickly (they have compliance teams and technical infrastructure), while small and medium enterprises and government agencies lag substantially.

A critical finding: Malaysia's AI contribution to GDP is projected at RM 480 billion by 2030, creating strong economic incentives to deploy AI rapidly. Governance frameworks that are perceived as slowing deployment face institutional resistanceโ€”even when the governance mechanisms are nationally mandated. This tension between economic ambition and governance rigor is not unique to Malaysia; it appears in virtually every national AI governance effort.

Hakimi et al. (2025), with 3 citations, analyze the challenges developing countries face in creating AI legal frameworks. Their analysis identifies a structural asymmetry: developing countries are rapidly adopting AI systems designed and trained in high-income countries, but they lack the regulatory infrastructure, technical expertise, and institutional capacity to evaluate and govern these systems.

The paper argues for a rights-based approachโ€”anchoring AI governance in existing human rights frameworks (right to privacy, right to non-discrimination, right to due process) rather than building AI-specific legal structures from scratch. This approach has practical advantages: human rights legal infrastructure already exists in most developing countries, and rights-based arguments carry political legitimacy that technocratic governance frameworks may lack.

Educational AI Governance

Rao (2025) develops a governance framework specifically for student use of generative AI tools (ChatGPT, Gemini, DALL-E) in educational institutions. The framework addresses a governance gap: most AI governance discussions focus on organizational deployment of AI systems, but students are using consumer AI tools in ways that fall outside institutional AI governance entirely.

The framework proposes three pillars: academic integrity (defining acceptable and unacceptable AI use in coursework), digital literacy (teaching students to evaluate AI outputs critically), and ethical reasoning (helping students understand the societal implications of AI systems they use daily). The argument is that AI governance in education should be educationalโ€”teaching governance principles alongside enforcing compliance rules.

AI Governance Landscape Comparison

<
JurisdictionApproachKey FeatureChallenge
EURisk-based regulation (AI Act)Mandatory compliance, heavy penaltiesCompliance burden on SMEs
USSector-specific, voluntary frameworksExecutive order + agency guidanceFragmented, limited enforcement
ChinaState-directed, content-focusedAlgorithm registration, content reviewTransparency concerns
MalaysiaPrinciple-based national frameworkSeven-principle governance modelImplementation gap in SMEs/government
Developing CountriesRights-based adaptationLeverages existing human rights lawTechnical capacity gaps
Educational InstitutionsPolicy + pedagogy hybridAcademic integrity + digital literacyStudent AI use outpaces policy

What To Watch

The global AI governance landscape is fragmenting along at least three axes: regulatory stringency (EU strict vs. US light-touch), political values (democratic vs. authoritarian AI governance), and economic development (capacity-rich vs. capacity-constrained). Watch for whether interoperability emergesโ€”mutual recognition agreements that allow AI systems certified in one jurisdiction to operate in anotherโ€”or whether fragmentation creates compliance nightmares for organizations operating across borders. The educational governance space will become increasingly important as AI-native students enter the workforce with habits formed by unregulated consumer AI use. Expect data ethics to become a core competency requirement, not a specialization, across all professions that handle personal dataโ€”which is to say, almost all professions.

Explore related work through ORAA ResearchBrain.

References (5)

[1] Britchenko, I., & Lysiak, I. (2025). EU Data Governance, AI Ethics, and Responsible Digitalisation in Higher Education. Digital Governance and Technologies Review, 4, 12-19.
[2] Mat Taib, S., Mohd Nazif, N.N.N., & Ariffin, A.S. (2025). Towards a Responsible AI Governance Frameworkโ€”Lessons from Policy Implementation in Malaysia.
[3] Hakimi, M., Zarinkhail, S., & Sahnosh, F.A. (2025). AI and Legal Reform in Developing Countries. Telematika: Jurnal Informatika dan Teknologi Informasi, 8(2).
[4] Rao, S. (2025). Establishing Responsible AI Use Policies for Students in Educational Institutions: A Framework for Governance, Ethics, and Innovation. Journal of New Zealand Educational Research.
Hakimi, M., Zarinkhail, S., & Sahnosh, F. A. (2025). Artificial Intelligence and Legal Reform in Developing Countries: Advancing Ethical, Rights-Based, and Accountable Digital Governance. Jurnal Ilmiah Telsinas Elektro, Sipil dan Teknik Informasi, 8(2), 127-144.

Explore this topic deeper

Search 290M+ papers, detect research gaps, and find what hasn't been studied yet.

Click to remove unwanted keywords

Search 7 keywords โ†’