Resources

International efforts to establish guardrails for AI by promoting safety, transparency, accountability, fairness, privacy, human oversight, and protection of human rights, with the EU AI Act being legally enforceable, while the OECD, UNESCO, and G7 frameworks provide voluntary guidance and best practices.

EU AI Act is a legally binding regulation that classifies AI systems across risk tiers: unacceptable, high, limited, and minimal risk, with explicit prohibitions on certain high-risk uses. Expected to take full effect by 2026, it is likely to serve as a global benchmark for AI governance. | More

OECD AI Principles were first established in 2019 and updated in 2024, encouraging governments to regularly review and adapt their AI-related policies. Broadly adopted, especially in OECD member countries. | More

UNESCO Recommendation on the Ethics of AI, the first global standard on AI ethics, voluntarily adopted by UN member states, promotes inclusive, sustainable, and ethical AI aligned with peace and human rights. | More

United Kingdom supports a ‘pro-innovation’, principles-based, sector-specific framework rooted in its 2023 AI Regulation White Paper, with oversight from regulators like the FCA and the Information Commissioner’s Office (ICO). | More

United States aligns to a fragmented approach, with state-level efforts including Colorado’s AI Act targeting bias in high-risk systems, Illinois’ HB 3773 addressing employment discrimination, and California’s SB-942 mandating transparency for generative AI.

China is focused on frameworks, including the Interim Measures for Generative AI Services and 2023 Deep Synthesis Measures, to address algorithmic accountability and deepfake risks.

Canada pursues principle-based oversight rooted in responsibility, transparency, and human-centric design, leveraging sectoral regulators and voluntary international standards; the Consumer Privacy Protection Act (CPPA) remains the cornerstone of privacy modernization effort.

The ISO/IEC 42001 is an AI Management Systems standard; widely adopted voluntarily alongside OECD AI Principles.

• EU AI Office is an EU body overseeing AI Act enforcement
• UK AI Safety Institute established after the 2023 AI Safety Summit, underscoring global safety leadership.
• US NIST (National Institute of Standards and Technology) publishes the AI Risk Management Framework (AI RMF)
• FCA (Financial Conduct Authority, UK) enforces ethical oversight in AI-driven finance.
• ICO (Information Commissioner’s Office, UK) ensures transparency via AI auditing.
• DARPA, the US Defense Advanced Research Projects Agency, promotes explainable AI and AI research.

• AI Now Institute (NYU) — focuses on the social implications of AI and policy research in responsible AI; research areas include algorithmic accountability, antitrust concerns, biometrics, worker data rights, and privacy. TechTarget
• Alan Turing Institute — committed to using data science and AI for everyone’s benefit and to protect society against unintended consequences.
• Center for Human-Compatible AI (CHAI) — a cooperation of universities to promote trustworthy AI and provably beneficial systems.
• Stanford HAI (Human-Centered AI Institute) — home of HELM and the AI Index Report
• Partnership on AI, a nonprofit of academic, civil society, industry, and media organizations exploring how to build AI systems that benefit all stakeholders; founding members include Amazon, Facebook, Google, Microsoft, and IBM. Created the AI Incident Database.
• International Association for Safe and Ethical AI (IASEAI), founded in 2024 and headquartered in Paris, focused on policy development, research, education, and community-building around AI safety.
• IEET (Institute for Ethics and Emerging Technologies) a think tank promoting how technological progress can increase freedom and human flourishing in democratic societies.
• AlgorithmWatch a research and advocacy organisation that evaluates algorithmic decision-making processes with social relevance.