2026年全球人工智能监管态势报告_39页_1mb

Summary of AI Regulation in 2026

Core Content

In 2026, AI regulation continues to evolve, with a focus on specific sectors and use cases. The year has seen a mix of tightening and softening of regulatory approaches, particularly in the United States and the European Union. While the US and EU have made efforts to reduce the strictness of AI regulations, they remain active in addressing key concerns such as bias, transparency, and the ethical use of AI in sensitive areas like employment, finance, insurance, and dynamic pricing. International cooperation is also on the rise, with global bodies like UNESCO and the UN playing a growing role in shaping AI governance.

Main Sectors and Key Laws

1. HR Tech Regulation

• Video Interview Laws: Focus on consent and notification for AI-driven video interviews.

  • Illinois Artificial Intelligence Video Interview Act: Requires notification and candidate consent for AI-driven video interviews, and submission of demographic data.
  • Maryland's Facial Recognition Services Prohibition: Prohibits facial recognition use in video interviews unless a waiver is signed.

• Bias Audit Laws: Mandate annual, independent audits to ensure fairness in automated employment decision tools (AEDTs).

  • New York City Local Law 144: Requires annual bias audits, results publication, and 10-day candidate notification.
  • New Jersey A3854 & A3855/S2964: Mandates bias audits, results publication, and 30-day or 10-day notifications for AEDT use.
  • New York A03914/S04394: Requires annual disparate impact analysis and results publication.

• Updates to Existing Frameworks: Extend equal opportunity protections to AI systems.

  • Illinois HB3773: Updates the Human Rights Act to include AI-based discrimination and zip code use as a proxy.
  • California's modified employment regulations: Extends anti-discrimination laws to AI systems and allows bias audits as a defense.

• Impact Assessment Laws: Require evaluations of AI systems' impact on employment and privacy.

  • New York A3779/S185: Mandates third-party impact assessments and prohibits reliance on AI alone for employment decisions.
  • Vermont H.262: Requires written impact assessments, restricts electronic monitoring, and bans certain biometric technologies.
  • Massachusetts S35 & Washington HB1672: Similar to Vermont, these laws require impact assessments and limit electronic monitoring.

• Worker Displacement and Wage Laws: Introduce financial deterrents to prevent AI-induced job displacement.

  • New York Workforce Stabilization Act: Requires bi-annual impact assessments and 2% surcharges for large-scale layoffs.
  • New York Robot Tax Act: Imposes taxes on displaced workers due to AI.
  • Illinois Surveillance-Based Price and Wage Discrimination Act & Georgia SB164: Prohibit using surveillance data for individual pricing and wages.

2. Financial Services Regulation

• AI is primarily regulated through testing environments and compliance frameworks.
  • UK FCA AI Live Testing: Supports AI deployment in financial markets with technical and regulatory assistance.
  • US Regulatory Sandboxes (S2528/HR4801): Allow AI experimentation without enforcement risk.
  • Singapore's AI Risk Management Guidelines: Require AI inventories, risk assessments, and lifecycle controls.
  • European Parliament Resolution: Calls for AI governance aligned with existing financial regulations and continuous monitoring.

3. AI in Insurance

• Several states have introduced laws to regulate AI in insurance practices.
  • Colorado SB169: Prohibits AI from making medical care decisions and requires annual reports.
  • Ohio HB579: Mandates annual AI use reports from health insurers and prohibits AI-only medical decisions.
  • Pennsylvania HB1925: Requires AI use disclosure and bias minimization attestation from insurers and healthcare providers.
  • Texas SB815: Prohibits AI in adverse healthcare determinations and allows commissioner audits.

4. Dynamic Price Setting Regulation

• Dynamic pricing, particularly in housing and retail, is under scrutiny.
  • New York S7882: Prohibits AI-driven rent setting algorithms, especially those using data from multiple landlords.
  • California AB 325: Bans "common pricing algorithms" that use competitor data to influence pricing or commercial terms.
  • Massachusetts S2515, Vermont H371, Illinois HB3838: Target dynamic pricing in grocery stores, retailers, and ticketing.

5. Generative AI Laws

• Generative AI is increasingly regulated, with a focus on transparency, safety, and ethical use.
  • California SB 53: Requires transparency reports, safety protocols, and catastrophic risk assessments from large AI developers.
  • New York RAISE Act: Mandates safety and security protocols, and imposes hefty penalties for non-compliance.
  • Maine LD 1727: Prohibits misleading users into thinking they are interacting with a human in AI chatbots.
  • Utah HB 452: Imposes restrictions on mental health chatbots, including no targeted advertising and no sharing of health data without consent.

6. Deepfake Laws

• Deepfake legislation targets non-consensual intimate imagery and impersonation.
  • US Federal TAKE IT DOWN Act: Prohibits publishing intimate visual depictions of minors or non-consenting adults and requires notice-and-action mechanisms.
  • Tennessee ELVIS Act: Protects artists from unauthorized use of their likeness or voice in AI-generated content.
  • Washington Forged Digital Likeness Law: Criminalizes the use of synthetic media for fraudulent or harmful purposes.

7. Risk-Based AI Frameworks

• Risk-based laws categorize AI applications based on their potential impact.
  • EU AI Act: The most comprehensive risk-based framework, influencing global standards.
  • Italy, Korea, Vietnam, Brazil, Chile, Colorado, California, Hawaii, Illinois: All have introduced or are developing risk-based AI laws.

8. International AI Initiatives

• Global cooperation is increasing, with regional and international bodies leading efforts.
  • UN Mechanisms: Promote multilateral AI governance.
  • UNESCO Guidance on Generative AI in Education: Focuses on ethical use in educational settings.
  • ASEAN Expanded Guide: Encourages AI governance and ethics across Southeast Asia.
  • Council of Europe Framework Convention on AI: Sets regional standards for AI development and use.

Key Takeaways

• Proactive Governance is Essential: With the AI regulatory ecosystem constantly evolving, businesses must proactively assess and manage their AI systems throughout the development lifecycle.
• Sector-Specific Focus: 2026 regulations are increasingly tailored to specific sectors (HR, finance, insurance, etc.), rather than being country-based.
• Global Coordination: International bodies are playing a more prominent role in shaping AI governance, with a growing emphasis on cross-border cooperation.
• Transparency and Accountability: Laws emphasize the need for transparency, disclosure, and accountability in AI systems, especially in high-risk areas like hiring, pricing, and mental health.
• Legal Penalties for Non-Compliance: Heavy fines and potential legal action are being used to enforce compliance with AI regulations, especially in cases involving bias, privacy, and safety.

Conclusion

The year 2026 marks a pivotal moment in AI regulation, characterized by a blend of sector-specific laws, international collaboration, and a continued emphasis on transparency, fairness, and safety. As AI continues to permeate various industries, regulatory frameworks are becoming more sophisticated and targeted, requiring businesses to be vigilant and proactive in their AI governance strategies.