Future Of AI Regulations: Global regulatory trends for AI

by Dr. S.RADHA RAMMOHAN, FIE
Professor & Program Head,
CODE – Center for Open & Digital Education
Chennai, India
Virtually delivered at the LMGlobal.Org‘s panel on November 15, 2025.

Presentation (PDF)

Abstract
[LMGlobal.Org] The United States is at an inflection point in governing artificial intelligence. At the federal level, the policy center of gravity has shifted within a short span: the Biden Administration’s 2023 executive order, which pushed safety testing and cross-agency guardrails, was revoked in January 2025, signaling a pivot from precaution to acceleration. The net effect is a moving target for developers, users, and the public. Any durable U.S. approach has to reconcile two facts: innovation leadership requires speed, while trust and adoption require credible accountability.

Even with that revocation, federal levers did not vanish. OMB’s 2024 guidance still set expectations for how agencies acquire and use AI, especially when public rights or safety are at stake. NIST remains the technical backbone through its AI Risk Management Framework and the 2024 Generative AI Profile, which many enterprises now treat as a de facto standard. These tools are voluntary, but they map cleanly to a risk-based approach and are likely to anchor any future federal baseline.

States are not waiting. Colorado enacted the first comprehensive, risk-based AI law in 2024 and has since delayed implementation to 2026 while revising scope and obligations. This state-led momentum, combined with divergent sector rules in privacy, employment, health, and finance, is creating a growing patchwork that will raise costs for national deployments and complicate compliance for startups and incumbents alike. A federal floor that preempts only where necessary, while preserving targeted state protections, is the pragmatic path to coherence.

Regulators also have existing powers. The FTC, DOJ, CFPB, and EEOC have made it explicit that there is no AI exemption in current law. That means unfair or deceptive practices, discrimination, and other statutory violations remain fully in scope, regardless of model architecture. Expect continued enforcement on claims, bias, security, data provenance, and dark patterns while Congress debates new statutes.

This chapter argues for a U.S. framework that is lean, risk-tiered, and interoperable with global regimes, and that uses enforcement and procurement as near-term stabilizers. The blueprint has five pillars: (1) a federal baseline that targets high-risk uses with duty-of-care, impact assessments, and incident reporting; (2) clear obligations for model and application layers, including secure-by-design and evaluations tied to context of use; (3) responsible disclosure and watermarking for synthetic media where material to consumer decisions; (4) federal procurement that requires NIST-aligned risk management as a condition of contract; and (5) agile oversight that funds testbeds, sandboxes, red-teaming, and post-deployment monitoring. With these moves, the U.S. can reduce regulatory uncertainty, protect people, and keep its innovation edge. The alternative is a costly patchwork that undermines both safety and competitiveness.

* * *

India–US AI Regulatory Approaches and the Future of Global AI Governance

India–US Comparative Overview

India uses a centralized, innovation-friendly model led by MeitY, supported by the DPDP Act 2023, synthetic content labeling, sector rules in finance and health, and a light-touch startup approach.
The US began with voluntary commitments but is moving toward binding laws due to concerns over bias, misuse, and national security.
Both nations are shifting from soft guidelines to enforceable regulations.

Global Shift to Risk-Based Regulation

High-risk AI in healthcare, hiring, policing, and critical infrastructure will face strict audits, while low-risk AI like chatbots will see lighter oversight.
This model promotes innovation with safety.
India, Japan, Canada, the EU, the UK, and the US are converging on this framework.

Transparency and Disclosure

Regulators seek clarity on datasets, limitations, and risks.
Model cards, system cards, impact assessments, watermarking, and AI labeling are becoming standard.

Sensitive Sector Regulation 
Elections
Controls on deepfakes and misinformation.
Healthcare: Accuracy testing and human oversight.
Criminal justice: Scrutiny of predictive tools and facial recognition.
Defense: Ethical norms for autonomous and military AI.

Liability and Accountability

Responsibility may fall on developers, users, or platforms depending on the harm.
Insurance and certification systems are expected to emerge.

Global Cooperation

UN, OECD, G7, and G20 are advancing shared AI safety and cybersecurity standards.
Future agreements may resemble climate treaties but with stronger security emphasis.

Data Privacy and Rights

Restrictions on data scraping, facial recognition, and biometric use are increasing.
Individuals may gain rights to explanation, opt-out, and correction.
The US may move closer to GDPR-like protections.

Open-Source vs. Closed-Source Models

Open-source enhances transparency but risks misuse; closed-source limits transparency and concentrates power. 
Future rules will assess capabilities and risks rather than licensing alone.

Conclusion

India is shaping a global move toward transparent, accountable, and risk-based AI governance that balances innovation and safety.
Global solutions is the only one solution and We need to create AI Self Regulatory system for various sectors along with AI regulations which is the Process of implementation.

Presentation (PDF)