Navigating the Future: Understanding the USA Administration’s Landmark Executive Order on AI

Introducing Our Author

Susan Palm, Chief Revenue Officer at 4CRisk.ai, has over 30 years leading teams focused on Compliance, Regulatory Affairs, InfoSec, Risk Management, Sales, and Customer Success. In this blog she shares her extensive experience helping companies stay ahead of risk events and regulatory changes in the USA based on the 2024 AI Executive order.

Summary of USA Administration’s Landmark Executive Order on AI

The executive order (EO) on artificial intelligence (AI) issued by the US Administration represents a first step in shaping the future landscape of AI. By implementing stringent transparency and safety testing requirements, the order aims to mitigate risks associated with powerful AI systems, ensuring they are secure and reliable before being widely adopted. While developers of advanced AI models must share their safety test results with the U.S. government, and companies are required to notify the federal government when training models that could pose significant risks, the EO, compared with the EU AI Act, may not raise the bar high enough, nor make consequences of non-compliance clear.  Businesses operating on a global scale will have significant compliance overlaps between the AI Act and EO requirements.

Sustainable and Trustworthy AI, in all its aspects, is needed now to build and maintain trust and accountability in AI technology. The EO emphasizes the importance of developing robust standards and tools to safeguard critical infrastructure and address potential threats in various domains, including cybersecurity and biological synthesis. The National Institute of Standards and Technology (NIST) will establish rigorous standards for extensive red-team testing, ensuring AI systems undergo thorough safety evaluations before public release. The Department of Homeland Security will apply these standards to critical infrastructure sectors and establish the AI Safety and Security Board, further bolstering national security measures.

This approach highlights the administration's commitment to advancing AI technology responsibly. By addressing both the risks and opportunities of AI, the EO strives to create a balanced environment where innovation can flourish while maintaining safety and security in a trustworthy ecosystem where AI can benefit society at large.

Still, many are asking if the EO goes far enough to temper the rapid rise of AI, especially in sectors beyond critical infrastructure.

What the Executive Order Covers

In addition to technical safeguards, the EO seeks to protect against the misuse of AI in life sciences by setting new standards for biological synthesis screening. Federal funding agencies will require these standards to fund life-science projects, creating powerful incentives for compliance and risk management.

The EO’s approach discusses eight main areas, such as national security, privacy, fairness, consumer protection rights, labor market practice, innovation potentiality, international cooperation, and governmental AI expertise. Special parts are dedicated to encouraging the ethical application of AI in the education sector, healthcare industry, and criminal justice system.

While some congressional hearings on AI have focused on the possibility of creating a new federal AI regulatory agency, today’s EO spreads out responsibility for AI governance among many federal agencies, tasking each with overseeing AI in their areas of expertise.

Comparing the Executive Order to the EU AI Act

The order's signing came at a critical juncture, coinciding with the G7's approval of the AI Code of Conduct and just days before an international summit on AI safety organized by the UK. This timing underscored the urgency of establishing a coherent national strategy on AI governance as world leaders prepare to discuss global security.  

The White House has released a fact sheet about the order:

• The order discusses civil rights, stating that the federal government must establish guidelines and training to prevent algorithmic bias—the phenomenon in which the use of AI tools in decision-making systems exacerbates discrimination.
• While the privacy provisions include some directives for federal agencies to strengthen their privacy requirements and support privacy-preserving AI training techniques, they also include a call for action from Congress. The President “calls on Congress to pass bipartisan data privacy legislation to protect all Americans, especially kids,” the order states. Whether such legislation would be part of the AI-related legislation that Senator Chuck Schumer is working on remains to be seen.
• This EO also uses Presidential powers to mandate primary agencies and departments across sectors to develop common industry standards, guidelines, and regulations applicable to artificial intelligence development and usage. This is different from the EU’s proposed AI Act, which is meant to create a single regulatory framework covering the European Union under single horizontal regulation.
• Despite the EO’s sectoral approach, which might introduce variations in standards within different industries and jurisdictions, it allows for some flexibility. On the other hand, the EU AI Act will apply directly in all Member States except those related to specific issues provided for by the Act.
• Beyond general principles, EO goes further by requiring specific actions to share safety test results and other vital information with the U.S. government, notify the government if they’re training a foundation model that could pose serious risks, and share results of red team testing. By 2025, several federal agencies are expected to publish AI guidelines, while new advisory committees and reporting schemes will also be set up.

Looking Ahead - Stronger Regulations and Consequences are Needed

While The EO is a good starting point, it does not address all AI-related challenges and even falls short of the standards set forth by the EU AI Act. In this sense, it also calls upon Congress to pass data privacy legislation and AI because there is a need for more permanent legal frameworks besides an easily undone EO. The policy provides tight deadlines for various actions, but previous experiences with AI-focused executive orders suggest that full implementation may be delayed.

As AI develops rapidly, it may take considerable time before its effects are felt. Businesses operating globally will need to comply with both the AI Act and EO requirements. Nevertheless, there is a tendency among EU countries to have more substantial document-based compliance proof, while the US appears to, at times, allow alignment with industry norms to count.

These next few months and years will be critical in assessing how well such measures can take off and if they can match the unstoppable progress of AI technology. It’s still too early to tell how this regulation will ultimately shake out. One thing is for sure, though: This EO isn’t going to slow down the development of powerful AI nor substantially mitigate the risks posed to everyday consumers and even world governments.

We look forward to more clarity, at least parity with the EU AI Act, and specific consequences for those who cannot comply.

Check out these related blogs and resources  

• https://www.4crisk.ai/post/gdpr-and-ai-regulation-how-ai-powered-products-minimize-operational-stresses-on-ai-data-custody-and-governance
• ‍https://www.4crisk.ai/post/gdpr-and-ai-regulation-escalating-privacy-compliance-challenges-for-organizations
• ‍https://www.4crisk.ai/whitepapers/adopting-ai-strategy-governance-and-evaluation-best-practices