Latham & Watkins operates worldwide as a limited liability partnership organized under the laws of the State of Delaware (USA) with affiliated limited liability partnerships conducting the practice in France, Hong

Kong, Italy, Singapore, and the United Kingdom and as an affiliated partnership conducting the practice in Japan. Latham & Watkins operates in Israel through a limited liability company, in South Korea as a

Foreign Legal Consultant Office, and in Saudi Arabia through a limited liability company. © Copyright 2023 Latham & Watkins. All Rights Reserved. Under New York’s Code of Professional Responsibility, portions

of this communication contain attorney advertising. Prior results do not guarantee a similar outcome. Results depend upon a variety of factors unique to each representation. Please direct all inquiries regarding

our conduct under New York’s Disciplinary Rules to Latham & Watkins LLP, 1271 Avenue of the Americas, New York, NY 10020-1401, Phone: +1.212.906.1200.

Initial Analysis of Private Sector Implications

On October 30, 2023, President Biden issued a sweeping Executive Order on the Safe, Secure, and

Trustworthy Development and Use of Artificial Intelligence (with an accompanying Fact Sheet). The

Executive Order appears designed to balance the goals of fostering innovation and remaining globally

competitive, with intense pressure to secure and regulate AI in the United States. It is structured around

eight “guiding principles and priorities”:

Supporting Workers: Federal agencies associated with labor and the workforce, such as the

Department of Labor, must develop principles and best practices to mitigate the harms and maximize

the benefits of AI for workers.

Advancing Equity and Civil Rights: Federal agencies across the government are directed to develop

new guidance, plans, and other measures within the scope of their authorities to combat the potential

for discrimination and other harms that AI may exacerbate.

Protecting Consumers, Patients, Passengers, and Students: Various agencies must ensure that AI is

Advancing Federal Government Use of AI: The federal government must develop guidance for

agencies’ use of AI, acquire AI products and services, and hire AI professionals.

laws to address risks that arise with use of AI, and to engage in further rulemaking specific to AI. The

Order also creates a new White House AI Council, with representatives of 29 federal divisions and

agencies to coordinate on implementation progress and efforts.

For companies and individuals, there are three overarching points:

First

viewing large AI models and the large computing clusters capable of use in their development and

training as matters of national security. The Order invokes the Defense Production Act (DPA) (50 U.S.C.

location of each computer cluster and the amount of total computing power available in each cluster.

These reporting requirements are the most specific new rules that come out of the Executive Order.

However, as of today, they apply only to companies whose models meet the definition of “dual-use

2

or that possess or offer computing centers

meeting certain thresholds.

3

Going forward, the Secretary of Commerce will define and update the set of

technical conditions for models and computing clusters subject to these reporting requirements.

Second

, the Order calls on federal regulators to regulate aggressively the use of AI in their respective

Third

, the requirements for AI will remain dynamic. The Order calls specifically for government studies as

well as future regulations across a wide swath of the federal government. Notable provisions include

requirements for:

The National Institute of Standards and Technology to develop — as the Fact Sheet describes the

requirement — “rigorous standards” for “extensive red-team safety testing to ensure safety before

public release” (Section 4.1(a)(ii))

The FTC to use its rulemaking authority “to ensure fair competition in the AI marketplace and to

ensure that consumers and workers are protected from harms that may be enabled by the use of AI”

(Section 5.3(a))

The Secretary of Commerce to issue regulations concerning reporting requirements related to the use

of certain AI models by foreign persons (Section 4.2(c), (d))

Order that will apply directly to businesses and provide initial analysis of the Executive Order.

Potential Impact on Businesses

circumstances as requiring telecommunications companies to provide information about their networks,

counter spying, and mandating that General Motors make ventilators during the COVID-19 pandemic.

As detailed below, the Order invokes the DPA to mandate that certain companies share particular

information with the government, including information related to model training, the ownership and

possession of model weights and physical and cybersecurity measures taken to protect such weights,

red-team testing results, and location and power of computing clusters. Notably, the White House has

relied on emergency powers under the DPA to impose these requirements, signaling that the White

House views large AI models as a national security issue. In addition to specific reporting requirements,

developers working on such models generally will need to implement security controls of the highest

degree to protect against foreign infiltration.

potential large-scale computing cluster” used to train AI.

Notably, many current AI models and computer clusters may not meet the thresholds of these

requirements. The Order sets forth initial thresholds for models and computing clusters, and gives the

Department of Commerce the authority to define the kinds of models and computer clusters that are

subject to these reporting requirements (Order section 4.2(b)). Until the Secretary of Commerce defines

those conditions, the initial thresholds are:

Models: Any model that was trained using a quantity of computing power greater than 10

integer or

floating-point operations, or using primarily biological sequence data and using a quantity of

Computing clusters: Any computing cluster that has a set of machines physically co-located in a

single datacenter, is transitively connected by data center networking of over 100 Gbit/s, and has a

Any ongoing or planned activities related to training, developing, or producing dual-use foundation

models, including the physical and cybersecurity protections taken to protect the training process

against sophisticated threats.

The results of any dual-use foundation model’s red-team testing based on guidance developed by

(including research and discovery, drug and device safety, healthcare delivery and financing, and

public health) (section 8(b)).

authority to require the entities it regulates to use appropriate AI methodologies to ensure compliance

with federal law, including to minimize bias or disparities for protected groups (section 7.3(b)).

in education,” including an “AI toolkit” for education leaders (section 8(d)).

assessing the need for information, technical assistance, and guidance related to AI (section 8(c)).

Secretary of Housing and Urban Development (HUD): The Order identifies a concern with AI

discrimination and biases in “access to housing and in other real estate-related transactions,” and

requests that HUD issue guidance on how AI may violate federal law and best practices companies

center networking of over 100 Gbit/s, and having a theoretical maximum compute capacity of 10

or floating-point operations per second for training AI.

with them to train a large AI model, by reporting to the government the identity of the foreign person, the

secure development practices for generative AI and for dual-use foundation models; and

Launching an initiative to create guidance and benchmarks for evaluating and auditing AI capabilities,

with a focus on capabilities through which AI could cause harm, such as in cybersecurity and

biosecurity.

NIST must establish guidelines to enable AI developers “to conduct red-teaming tests to enable

deployment of safe, secure, and trustworthy systems,” including:

Developing guidelines to assess and manage the safety, security, and trustworthiness of dual-use

foundation models; and

: After gathering input from the private sector and other stakeholders

via a public consultation process, the Secretary of Commerce shall submit a report to the President on

provide input into this process, given that the Secretary is tasked expressly with soliciting input from the

private sector.

Practical Takeaways

The Executive Order is a significant step towards furthering the governance of AI in the United States and

beyond. And the timing of the Executive Order’s issuance is notable. It was issued contemporaneously

with the Hiroshima AI Process Comprehensive Policy Framework—a set of international guiding

principles on AI and a voluntary code of conduct agreed by the leaders of the G7—also published on

October 30, 2023 (G7 Guiding Principles). It was also published just days before the AI Safety Summit,

hosted by the UK Government, at which key figures in the AI industry were to gather how to best manage

If you have questions about this Client Alert, please contact one of the authors listed below or the Latham

lawyer with whom you normally consult:

Michael H. Rubin

Andrew Gass

Client Alert is published by Latham & Watkins as a news reporting service to clients and other friends.

The information contained in this publication should not be construed as legal advice. Should further

analysis or explanation of the subject matter be required, please contact the lawyer with whom you

normally consult. The invitation to contact is not a solicitation for legal work under the laws of any

jurisdiction in which Latham lawyers are not authorized to practice. A complete list of Latham’s Client

Alerts can be found at www.lw.com. If you wish to update your contact details or customize the

information you receive from Latham, visit our subscriber page.

2

4

5

6

7

9

Service Product. “Infrastructure as a Service Product” means any product or service offered to a consumer, including