The rapid advancement of Artificial Intelligence (AI), particularly the emergence of powerful General-Purpose AI (GPAI) models, has necessitated a swift and significant evolution in governance frameworks worldwide. These frameworks are no longer theoretical constructs but are transforming into concrete, actionable policies designed to ensure responsible development and deployment. From international collaborations to regional legislative acts, the global community is grappling with how to harness AI’s potential while mitigating its inherent risks. The landscape of AI governance is dynamic, with new regulations and initiatives constantly emerging to keep pace with technological innovation.

The Global Push for Harmonized AI Governance

The inherently global nature of AI development and deployment demands international cooperation. Several key organizations and initiatives are leading the charge in establishing foundational principles and fostering cross-border dialogue, aiming for a more unified approach to AI regulation, according to Bradley.

• OECD AI Principles: Adopted in 2019 and updated in 2024, these principles serve as a cornerstone for many national AI strategies. They emphasize a rights-based approach, guiding the development and deployment of AI systems to promote human rights and democratic values. Governments globally utilize these recommendations to design policies and risk management frameworks, laying the groundwork for global interoperability across regulatory jurisdictions, as highlighted by EvalCommunity.

• UNESCO Recommendation on AI Ethics: This framework provides a globally accepted normative standard, embedding human rights, dignity, inclusion, fairness, non-discrimination, social justice, and environmental well-being into AI development and deployment, according to Bradley.

• UN Initiatives: The United Nations has established several bodies to address AI governance. The High-Level Advisory Body on AI analyzes the current situation and recommends strategies for international governance. The Global Digital Compact, adopted in September 2024, is a comprehensive framework for global cooperation in governing digital technology and AI, prioritizing human rights and security, as detailed by UN.org. Furthermore, the Global Dialogue on AI Governance and the Independent International Scientific Panel on AI provide platforms for multi-stakeholder discussions and evidence-based insights into AI’s opportunities and risks, according to UN.org and WEF.

• AI Safety Summits: Events like the Bletchley Park Summit (2023), Seoul Summit (2024), and Paris AI Action Summit (2025) have brought together international governments, industry leaders, and civil society to focus on the risks posed by “frontier AI” – highly capable general-purpose AI technologies. The Bletchley Declaration, for instance, calls for international cooperation to manage AI safety risks, as reported by Freshfields. These summits aim to build a shared understanding of risks, foster international collaboration on safety research, and develop appropriate measures for organizations, according to Gov.uk.

Regional Regulations Taking Shape: The EU AI Act and US Approaches

While global principles provide a foundation, regional and national regulations are translating these into legally binding requirements and practical guidelines, creating a patchwork of compliance obligations for developers and deployers.

The European Union’s Pioneering AI Act

The EU AI Act, which entered into force on August 1, 2024, stands as the world’s first comprehensive legal framework for AI systems, according to Wikipedia. Its provisions will gradually come into operation over the following 6 to 36 months. The Act adopts a risk-based approach, categorizing AI applications into four levels: unacceptable, high, limited, and minimal risk, as explained by Nemko.

For General-Purpose AI (GPAI) models, the Act imposes specific transparency requirements, with additional evaluations for high-capability models. GPAI models presenting systemic risk—defined, for example, by a cumulative amount of compute used for training greater than 10^25 FLOPs—face even stricter obligations, according to Orrick. These include:

• Thorough risk assessments and mitigation strategies.
• Incident reporting to the AI Office.
• Ensuring robust cybersecurity measures.
• Maintaining detailed technical documentation and providing comprehensive information to downstream deployers, as outlined by TestingXperts.

To facilitate compliance, a Code of Practice for GPAI models was developed by the EU AI Office and various stakeholders, published in July 2025, according to ArtificialIntelligenceAct.eu. Member States are also establishing regulatory sandboxes to allow developers to trial AI systems under supervision, fostering innovation while maintaining safeguards.

The United States’ Evolving Stance

The US approach to AI governance has seen shifts and developments across different administrations, often characterized by a sector-specific or voluntary framework approach, as noted by Wikipedia.

• NIST AI Risk Management Framework (AI RMF 1.0): Released in January 2023, this is a voluntary set of guidelines designed to help organizations incorporate trustworthiness into the design, development, use, and evaluation of AI systems. It outlines seven key characteristics of trustworthy AI: validity and reliability, safety, security and resilience, accountability and transparency, explainability and interpretability, privacy enhancement, and fairness with bias management, according to Verifywise.

• Executive Orders (EOs):

  • The Biden administration’s Executive Order 14110 (October 2023) focused on AI safety and security.
  • The Trump administration, in January 2025, revoked previous AI policies and issued EO 14179, aiming to remove barriers to American AI leadership, as reported by WSGR.
  • Subsequent EOs, such as EO 14365 (December 2025) and the June 2026 EO “Promoting Advanced Artificial Intelligence Innovation and Security,” have increasingly emphasized national security and cybersecurity risks, according to Skadden. The June 2026 EO directs the creation of a framework for developers of advanced frontier models to engage in voluntary pre-release review with the federal government.

• State-level regulations: Several US states, including California, have enacted their own AI-related laws, creating a complex, multi-layered regulatory landscape that businesses must navigate.

Key Trends Shaping the Future of AI Governance

The evolution of AI governance frameworks highlights several critical trends that will define the future of responsible AI development and deployment, according to ResearchGate.

• Shift from Principles to Practicality: There’s a clear movement from abstract ethical guidelines towards concrete, enforceable policy mechanisms and technical standards. This includes the development of technical tools for bias detection, explainability, and risk assessment, as noted by Tetrate.io.

• Risk-Based and Iterative Approaches: Regulators are increasingly adopting risk-based frameworks to tailor obligations to the potential impact of AI systems. Furthermore, the dynamic nature of AI necessitates iterative regulatory models that can adapt to rapidly evolving technology, ensuring that governance remains relevant and effective.

• Emphasis on Trustworthiness: Core principles like transparency, accountability, fairness, human oversight, and data protection are consistently at the forefront of governance discussions. Organizations with robust AI governance are better positioned to build trust with stakeholders and reduce legal and reputational risks, according to Databricks.

• Focus on General-Purpose and Frontier AI: The unique characteristics of GPAI models, with their broad applicability and potential for systemic risks, are driving specific regulatory attention and requirements, as discussed at the UK AI safety summit.

• International Harmonization: The global nature of AI development underscores the need for international cooperation and interoperability among different regulatory approaches to avoid fragmentation and ensure consistent standards, according to Databricks.

As AI continues to integrate into every aspect of life, the ongoing evolution of governance frameworks will be crucial for ensuring that these powerful technologies are developed and deployed responsibly, ethically, and for the benefit of all. The journey towards comprehensive and effective AI governance is complex, but the concerted efforts of governments, industry, and civil society worldwide are paving the way for a future where AI’s potential can be fully realized while its risks are carefully managed.

Explore Mixflow AI today and experience a seamless digital transformation.

Explore Mixflow AI today and experience a seamless digital transformation.

References:

• bradley.com
• evalcommunity.com
• un.org
• weforum.org
• un.org
• freshfields.com
• www.gov.uk
• wikipedia.org
• techuk.org
• wikipedia.org
• nemko.com
• verifywise.ai
• verifywise.ai
• orrick.com
• eipa.eu
• testingxperts.com
• artificialintelligenceact.eu
• artificialintelligenceact.eu
• snowflake.com
• wikipedia.org
• mcdermottlaw.com
• wsgr.com
• skadden.com
• researchgate.net
• tetrate.io
• databricks.com
• databricks.com
• china-embassy.gov.cn
• UK AI safety summit general purpose AI