This fourth and last part of our breakdown of the AI Executive Order covers the remaining sections 9 through 12 (leaving aside the boilerplate section 13 - general provisions). Here are the key takeaways:
Section 9 - Protecting Privacy
The order directs a comprehensive assessment of personal information within commercially available information, as well as guidelines and safeguards to protect the privacy of individuals.
The National Science Foundation is directed to integrate privacy-enhancing technologies into their operations.
Section 10 - Advancing Federal Government Use of AI
The use of AI by federal agencies and by the federal government will be coordinated. The Executive Order lays out the governance structure for this.
For the federal use of AI, recommendations for AI testing, including external testing and red-teaming, watermarking, risk management, reporting, and continuous monitoring will be published.
The Executive Order does not foresee a general ban on government use of generative AI, but restrictions that allow the use of low-risk AI services.
Recommendations and programs to attract and hire AI talent for the federal government and agencies will be established.
For civil engineers concerned with critical infrastructure, the set of required AI skills will be defined.
Section 11 - Strengthening American Leadership Abroad
International collaboration, also with the goal to advance the partners' understanding of U.S. AI policies, will be enhanced.
The U.S. will work to develop voluntary international support of the U.S. policies and restrictions with regard to AI.
The development of global technical standards for AI development and use, based on the NIST AI Risk Management Framework and the United States Government National Standards Strategy for Critical and Emerging Technology will be advanced.
An AI in Global Development Playbook will be published that incorporates the AI Risk Management Framework’s principles, guidelines, and best practices into the social, technical, economic, governance, human rights, and security conditions of contexts beyond U.S. borders.
International collaboration on the protection of and recovery from interruptions of critical infrastructure will be driven.
Section 12 - Implementation
To drive and oversee the implementation of the Executive Order, a White House AI Council has been established, chaired by the Deputy Chief of Staff for Policy and including 29 Department Secretaries and federal agency heads
Conclusion of this mini-series:
The AI executive order lays out an exhaustive framework for regulatory activities in all relevant technological and societal domains. It is very detailed with regard to governance structures and timelines. The urgency of the Executive Order becomes evident in the very ambitious timelines for implementation and the top-level oversight of the activities. The speed with which the U.S. has overtaken European AI regulation is remarkable and only possible by making this topic a Presidential priority and by bypassing the lawmaking process in Congress. A risk of this approach is that a future president will be able to revoke the AI Executive Order at any time, also without involvement of the Congress.
Dear reader, many thanks for following this mini-series. If you found it helpful, please don't hesitate to share it with your friends and colleagues. Stay tuned for more insights and thoughts on this channel.