Author	Niamh Libonatti-Roche
Date	28th November 2023

Key Takeaways: US Executive Order on AI

Executive Summary

During the Bletchley Conference on Artificial Intelligence, Vice President Kamala Harris announced the US strategy to AI. The key business takeaways from the Executive Order fall under the headings of:

• Improving AI Safety and Security
• Improving AI Deployment
• Creating AI Rights
• Auditing and improving Governmental use of AI
• Assuring Privacy Protections

This bulletin explains what the Executive Order proposes under these headings and provides a short business-focused analysis of the impact. That can be summarised by saying that the effect of the US Executive order on AI has a global impact on those developing, deploying, or using AI that processes US citizens’ data.

US Executive Order on AI teardown

This section summarises the US Executive Order on AI

Improving AI Safety and Security

• Developers of most powerful AI must undertake testing to ensure the safety of their models. They must also share safety test results with the US government.
• The US Government and relevant agencies will:
  • Develop standards, tools, and tests to ensure that AI systems are safe, secure, and trustworthy.
  • Develop strong new standards for biological synthesis screening, to protect against the risks of using AI to engineer dangerous biological materials.
  • Establish standards and best practices for detecting AI-generated content and authenticating official content to protect Americans from AI-enabled fraud and deception.
  • Establish an advanced cybersecurity program to push-forwards the development of AI tools that find and fix vulnerabilities in critical software, making US software and networks more secure.

Improving AI Deployment

The US Government will:

• Ensure the continued American collaboration with other international powers to ensure that AI deployment is safe, secure, and trustworthy.
• Create a fair, open, and competitive AI ecosystem and ensure that researchers, student, developers, and entrepreneurs have access to the necessary technical assistance and resources.
• Support workers by developing best practice to mitigate potential harm to workers whilst allowing them to harness the massive benefits that AI can offer.
• Provide resources to support the responsible use of AI in all settings with specific focus on advancing responsible use of AI in healthcare and drug development and assuring consumer rights.

Ensuring against discrimination and protecting AI and Civil Rights

The US Government will:

• Continue its work on establishing an AI Bill of Rights.
• Provide clear guidance to public sector bodies e.g., agencies and the private sector on mitigating algorithmic discrimination when developing, deploying or using AI.
• Address algorithmic discrimination through testing, training, technical assistance, and co-ordination between Department of Justice and Federal Civil Rights Office on best practices for investigating and prosecuting civil rights violations stemming from AI.

Auditing and improving Governmental use of AI

• Develop best practice standards for the use of AI in the criminal justice system.
• Order the development of a National Security memorandum that directs the US military, intelligence community to use AI safely, ethically, and effectively in their missions.

Assuring Privacy Protections

Government will:

• Work to better protect American’s Privacy through a bipartisan, national data privacy legislation that protects all Americans and measures to evaluate and improve how federal agencies, businesses and researchers use private data.
• Invest in research and technologies that preserve privacy.

EOAI Analysis

The US Executive Order on AI has made significant steps forward to creating a US regulatory framework for AI and has unexpectedly advanced the US approach to privacy regulation.

Definition of AI

The Executive Order uses an extremely broad definition of AI.

“a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments.”^[1]

That EO scope is not limited. The effect is that it applies to any machine-based system that makes predictions, recommendations, or decisions. This is a particularly broad definition and means many more systems will be covered by future requirements and recommendations on AI made by the US government.

The Executive Order also provides assurance for those developing AI that to have met the development standard in the US, companies must comply with the NIST AI Framework, as it is NIST and their framework that are identified as the source for development of key AI guidelines. As such companies who currently trade or who are based in the US and regularly develop AI models should look to pro-actively conform with this standard.

AI Rights Bill

The proposal of an AI Rights Bill is something that US companies involved in AI development or who regularly deploy AI should evaluate and try to align their practices to ensure that US citizens’ AI rights are protected.

Changed AI Landscape

The US Executive Order on AI is an extremely interesting development in the AI landscape. Because, much like the UK, the US approach is:

• largely “soft law”
• doesn’t create legally binding rules and regulations that effect AI companies and users

However, the fact that the AI approach is contained in an Executive Order means that the obligations imposed on certain agencies, and on the US government, must be acted upon. It:

• can act as a starting point for new laws and regulations and,
• suggests that the Biden/Harris administration is looking to begin this process.

Privacy Law Impact

Until now the US has resisted introducing a nationwide privacy legislation, leading the US Privacy Law landscape to become a patchwork of standards.

However, the Executive Order on AI has called for bi-partisan collaboration to  create and bring into force US-wide privacy legislation. This is a landmark moment in the world of Privacy.

Companies who trade with the US or use US citizens data should stay alert to progress on this new Privacy law as should the privacy professionals who support them. And those companies should also strive to understand the impact of processing huge volumes of data to train AI – especially if this data is personal or special category data.

---

[1] Found 15 U.S.C. 9401(3):

How can we help you?

IFor more information on international positions being taken on AI regulation, to seek our help to establish fair, transparent, and legal processing of personal data, or for access to our whitepapers get in touch by clicking here.