Current and Emerging AI Regulations

Here are statistics on AI security and safety regulations, including both existing and upcoming frameworks, with their sources and dates.

Existing and Recently Passed Regulations

• The EU AI Act, a landmark regulation for AI, was officially adopted on May 21, 2024. It categorizes AI systems by risk level and imposes strict requirements on high-risk systems, including cybersecurity provisions and a ban on certain applications.¹

• In the US, the National Institute of Standards and Technology (NIST) released its AI Risk Management Framework (AI RMF 1.0) in January 2023.² It provides voluntary guidance for organizations to manage the risks of AI.³
  
  
• The UK announced a £100 million commitment to establish an AI Safety Institute in November 2023to develop new tools for evaluating AI safety and security.

Upcoming Regulations and Legislative Efforts

• As of May 2024, 15 US states had enacted laws related to AI, and another 30 states had introduced or were considering legislation. A key focus of these bills is AI transparency and government use of AI.

• The G7 is working on a Code of Conduct for AI developers, a voluntary international framework focused on safe and responsible AI development.⁴ The code was initially proposed in October 2023.

• China is developing a comprehensive set of AI regulations, with draft rules on generative AI released in July 2023. These rules focus on content moderation, security, and a system for registering algorithms with the government.⁵

• In the US, there are several federal legislative proposals aimed at AI, including bills that would mandate risk assessments for high-risk AI systems and require transparency regarding the use of synthetic media.⁶ These bills remain under consideration as of mid-2024.

Governments and international organizations are rapidly increasing their efforts to regulate AI, with a significant surge in legislative activity over the past couple of years.¹ The focus has been on establishing new laws and frameworks to address the risks associated with AI development and deployment.²

Some Statistics on AI Regulations

• Growing Number of Laws: In 2024, U.S. state lawmakers introduced over 600 AI-related bills, with 99 of them enacted into law.³ This marked a significant jump from 2023, where fewer than 200 such bills were introduced. Similarly, U.S. federal agencies introduced 59 AI-related regulations in 2024, more than double the number from 2023.⁴ Globally, legislative mentions of AI have risen 21.3% across 75 countries since 2023.⁵

Organizations Involved in AI Regulation

A wide range of government bodies and international organizations are actively involved in creating AI regulations.⁶

United States:

• National Institute of Standards and Technology (NIST): Released the AI Risk Management Framework to provide guidance for managing AI risks.⁷
• 
  
• Small Business Administration (SBA) and National Science Foundation (NSF): They provide grants for AI R&D through programs like SBIR and STTR.
• Federal Trade Commission (FTC), Equal Employment Opportunity Commission (EEOC), and Department of Justice (DOJ): These agencies are applying their existing legal authority to AI-related issues, such as discrimination and consumer protection.⁸

International:

• The European Union (EU): The EU AI Act, adopted in May 2024, is the most comprehensive AI regulation to date, categorizing AI systems by risk level.⁹ The EU has also established new bodies like the EU AI Office to enforce these regulations.¹
  
• The G7: Working on a Code of Conduct for AI developers, a voluntary international framework for safe AI development.



• The United Nations (UN) and UNESCO: They have developed global standards and recommendations for AI ethics, such as UNESCO's "Recommendation on the Ethics of Artificial Intelligence."¹¹

• The Council of Europe: Developed a legally binding treaty on AI to safeguard human rights and the rule of law.¹²

Other Important Details

• Risk-Based Approach: Many recent regulations, including the EU AI Act, use a risk-based framework that imposes stricter rules on AI systems that pose a higher risk to human rights and safety.¹³ This approach is being adopted in other regions as well.¹⁴
  
  
• Voluntary vs. Mandatory: While some international frameworks, like the G7's Code of Conduct, are voluntary, there is a clear trend toward mandatory, legally binding regulations at the national and regional levels.

• Focus Areas: Recent legislation is often focused on specific harms, such as political deepfakes, non-consensual sexual imagery, and biometric data privacy.¹⁵ This shows a shift from broad principles to targeted rules.¹⁶
  
  
• Lack of Harmonization: The absence of a single federal framework in the US has led to a "patchwork" of laws at the state level, creating challenges for companies trying to comply with multiple, often conflicting, regulations.¹⁷

Sources and Dates

• EU AI Act: European Parliament and Council of the European Union, adopted May 21, 2024.⁷
  
• NIST AI Risk Management Framework: National Institute of Standards and Technology, released January 26, 2023.⁸
  
• UK AI Safety Institute: UK Government Announcement, November 1, 2023.⁹
  
• US State-Level Legislation: National Conference of State Legislatures (NCSL), data updated through May 2024.
• G7 AI Code of Conduct: G7 Leaders' Statement on AI, released October 30, 2023.
• China AI Regulations: Cyberspace Administration of China, draft rules published July 2023.

Based on novel mathematical understanding