AI governance is undergoing a significant reset. Colorado just replaced its comprehensive AI Act with a streamlined disclosure-based framework, while international efforts to train the next generation of AI policymakers are ramping up. These parallel developments reveal a critical tension: the world is racing to regulate AI, but there's still no consensus on how to do it effectively (Sources 1, 2).

What Happened to Colorado's AI Regulation?

On May 14, 2026, Colorado Governor Jared Polis signed SB 189, "Automated Decision-Making Technology," into law, repealing the state's original Colorado AI Act just weeks before it was set to take effect on June 30, 2026. The new law, which takes effect on January 1, 2027, represents a dramatic shift in regulatory philosophy.

The original Colorado AI Act required developers and deployers of high-risk AI systems to conduct risk assessments, perform impact assessments, and take reasonable care to prevent algorithmic discrimination. SB 189 eliminates these obligations entirely, replacing them with a lighter-touch disclosure framework focused on "automated decision-making technology" (ADMT) used in consequential decisions.

Under the new law, companies must provide consumer notices and post-adverse outcome disclosures, but the compliance burden is substantially reduced. The law applies to AI systems used to materially influence decisions in seven covered domains:

• Education: AI systems used in admissions, enrollment, or academic decisions
• Employment: Hiring, promotion, and termination decisions involving AI
• Financial and lending services: Credit decisions, loan approvals, and pricing
• Essential government services: Public benefits and welfare determinations
• Healthcare services: Treatment decisions and care access
• Residential real estate: Housing decisions and pricing
• Insurance: Coverage, pricing, and claims decisions

How Does Colorado's New Approach Compare to Other States?

Colorado's revised framework is notably different from how other states are approaching AI regulation. Most states, including California, Illinois, Texas, New York, and Utah, have adopted sector-specific or use-case-specific laws targeting narrow categories like employment screening or healthcare chatbots. Colorado's approach, even in its streamlined form, applies across multiple domains and governs both developers and deployers of AI systems.

The new Colorado law resembles California's recently finalized Consumer Privacy Act (CCPA) Automated Decision-Making Technology regulations. Both frameworks require upfront disclosure to consumers before or at the point of interaction with ADMT. However, they diverge in structure: SB 189 provides post-adverse outcome rights, allowing consumers to seek explanations and corrections after a negative decision, while the CCPA establishes proactive rights, including the ability to opt out of ADMT use regardless of outcome.

The Colorado Attorney General is directed to adopt rules prior to January 1, 2027, to clarify the content and format of post-adverse outcome disclosures and implement consumer rights provisions, including access, correction, and meaningful human review. These forthcoming rules are expected to provide critical details on how the law will operate in practice.

Why Is Global AI Governance Training Expanding?

While states like Colorado are debating the right regulatory approach, international institutions are preparing the next generation of AI policymakers. The AI Global Governance (AIGG) Summer School, now in its third edition, is being held in Brussels from May 26 to 30, 2026, to equip professionals with the knowledge and skills needed to navigate AI governance in an increasingly fragmented world.

The program examines how AI is transforming the global balance of power, reshaping cooperation among states, and changing how countries prepare for and conduct warfare. It also addresses emerging governance challenges such as algorithmic bias, digital authoritarianism, and the use of autonomous systems in conflict. Participants explore these issues through comparative perspectives from the European Union, the United States, Japan, and China, analyzing how AI affects international relations, human rights, labor markets, and environmental sustainability.

The summer school strengthens participants' ability to engage critically with the policy and institutional dimensions of AI governance. Through expert-led sessions, simulations, and institutional visits to the EU, NATO, and the United Nations Development Programme, participants learn to assess systemic risks, identify governance gaps, and design strategies to mitigate unintended consequences.

What Skills Are AI Governance Professionals Developing?

The AIGG Summer School emphasizes practical, forward-thinking skills that go beyond theoretical knowledge. Participants develop expertise in several key areas:

• Technology Awareness: Familiarity with key AI technologies, including autonomous systems, machine learning, and AI safety, to understand the technical challenges in governance
• Critical Thinking: Ability to assess and analyze complex AI governance issues from multiple perspectives, including ethical, legal, and societal dimensions
• Policy Analysis: Skill in evaluating and comparing regulatory frameworks and understanding their implications for global AI governance
• Ethical Reasoning: Ability to assess the ethical implications of AI technologies and their potential impact on society, with a focus on fairness, transparency, and accountability
• International Relations Knowledge: Understanding of global governance structures, international relations, and the role of multilateral organizations like the G7, OECD, and the UN in shaping AI policy
• Problem-Solving: Capability to develop practical, innovative solutions for complex AI governance challenges, particularly in the context of developing countries and emerging technologies
• Adaptability and Global Awareness: Capacity to adjust to rapidly evolving AI technologies and understand their impact across different regions and contexts

The program welcomes applicants from government, international organizations, think tanks, academia, civil society, and the private sector, including policymakers, diplomats, regulatory professionals, researchers, analysts, legal specialists, and technology experts.

How Should Companies Prepare for Colorado's New AI Regulation?

Developers and deployers of ADMT systems should take several concrete steps to prepare for compliance with SB 189 before the January 1, 2027 effective date:

• Identify Covered Use Cases: Map where ADMT is used to materially influence consequential decisions across the business, focusing on the seven covered domains
• Understand Liability Allocation: Recognize that both developers and deployers may be held liable for unlawful discrimination claims, with fault allocated based on each party's relative responsibility
• Prepare for Attorney General Guidance: Monitor forthcoming rules from the Colorado Attorney General on post-adverse outcome disclosure formats and consumer rights implementation
• Review Contractual Provisions: Ensure that indemnification clauses do not attempt to shield parties from liability for their own discriminatory conduct, as such provisions are void under the law
• Maintain Compliance Documentation: Document how ADMT systems are intended to be used and ensure that downstream uses align with documented purposes to limit developer liability exposure

The law clarifies that compliance with SB 189 does not provide a defense to liability under other applicable laws, including existing anti-discrimination and consumer protection statutes. This means companies must continue to meet all existing legal obligations while also adhering to the new disclosure and consumer rights requirements.

What Does This Mean for the Future of AI Governance?

Colorado's shift from a comprehensive framework to a disclosure-driven approach signals a broader debate about how to regulate AI effectively. The state's decision to repeal its original AI Act suggests that policymakers are still grappling with the right balance between innovation and oversight. Whether Colorado's revised approach becomes the prevailing model for broader state AI regulation remains uncertain, but it demonstrates that governance frameworks are evolving rapidly in response to industry feedback and practical implementation challenges.

Meanwhile, the expansion of international AI governance training through programs like the AIGG Summer School reflects recognition that AI regulation cannot be solved at the state or national level alone. As AI systems become increasingly global in scope, governance frameworks must account for cross-border implications, geopolitical competition, and the need for coordinated international standards. The tension between Colorado's lighter-touch approach and the comprehensive governance frameworks being studied internationally suggests that the next phase of AI regulation will involve ongoing negotiation between innovation and oversight, between national sovereignty and global coordination (Sources 1, 2).