Summary: What Does Biden’s Executive Order on Artificial Intelligence Actually Say?
November 7, 2023 |
Last week, the Biden-Harris Administration issued its Executive Order on AI: “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence.”
The EO emphasizes the need for regulation of high-risk AI and—critically—recognizes the link between privacy and AI. The order promises federal support for the development and use of privacy-enhancing technologies (PETs) and requires agencies to evaluate how they use commercially available information containing personally identifiable information, including data obtained from data brokers. In announcing the order, President Biden also called on Congress to pass bipartisan data privacy legislation.
In addition to calling out these privacy concerns, the order also recognized the civil rights violations associated with the use of AI and automated decision-making. The EO directs the Attorney General to take steps to mitigate algorithmic discrimination and other civil rights violations resulting from the use of AI and agencies to provide guidance to landlords and public benefits providers about keeping AI from exacerbating discrimination.
Perhaps the most tangible action in the EO is that the developers of certain AI systems are now required to share information with the government. The EO uses the Korean War-era Defense Production Act, an act typically invoked during national emergencies, to require that developers of powerful foundation models tell the government when they are training the models and share the results of all red-team safety tests.
Other technical directives include requiring NIST to set standards for pre-deployment red-team testing for AI foundation models and requiring the Department of Commerce to develop guidance around watermarking, content authentication, and other labeling tools that agencies will use to differentiate between authentic communications from the government and AI-created content.
Another key area of focus for the executive order was on the rapid hiring of AI professionals as well as AI training for current employees. The order directed agencies to develop a more efficient and faster process for AI procurement and to further study how to increase AI use within the federal government.
Here’s a detailed breakdown of what the EO requires.
The EO instructs several agencies to solicit public input, publish Requests for Information (RFI), or consider initiating rulemakings on various AI-related issues.
- Secretary of Commerce shall solicit input on potential risks and benefits of and on appropriate policy and regulatory approaches to dual-use foundation models for which model weights are widely available.
- Secretary of Labor shall publish an RFI to ID AI occupations for which there is an insufficient number of qualified US workers.
- OMB director shall issue an RFI to inform potential revisions to guidance to agencies on implementing privacy provisions of the E-Government Act.
- Secretary of State shall consider initiating rulemaking for: (1) new criteria to designate countries and skills on the Department of State’s Exchange Visitor Skills List as it relates to the 2-year foreign residence requirement for certain J-1 nonimmigrants and (2) to expand categories of nonimmigrants who qualify for the domestic visa renewal program to include academics and students.
Developing guidance and best practices
A large portion of the order is focused on directing agency heads to develop and publish guidance, best practices, or strategies about how AI should be used responsibly.
- OMB director shall: (1) develop a method for agencies to track and assess their ability to adopt AI; (2) develop an initial means to ensure agency contracts for AI acquisition align with guidance issued by the Directors of OMB and OSTP; and (3) issue guidance to agencies to strengthen effective and appropriate use of AI, advance AI innovation, and manage risks. (EPIC will post a more detailed analysis of the draft OMB guidance later.)
- Secretary of Commerce and the NIST Director shall: (1) establish guidelines and best practices for development and deployment of safe, secure, and trustworthy AI; (2) develop guidelines, tools, and practices to support implementation of minimum risk-management practices; and (3) create guidelines for agencies to evaluate the efficacy of differential privacy guarantee protections.
- Secretary of Commerce and the OMB Director shall develop guidance regarding existing tools for content authentication.
- Secretaries of Commerce and Homeland Security shall incorporate AI RMF guidance into safety and security guidelines for critical infrastructure owners and operators.
- Secretary of Energy shall: (1) develop and implement a plan for developing Energy’s AI model evaluation tools and AI testbeds, including guardrails to reduce risks, and (2) develop tools to facilitate building foundation models useful for basic and applied science.
- Assistant to President for National Security Affairs and the OMB director shall coordinate work by heads of agencies in charge of critical infrastructure to develop and take steps for the federal government to mandate guidelines.
- Chief Data Officer Council shall develop initial guidelines for performing security reviews while also providing public access to federal government data.
- Under Secretary of Commerce for IP and USPTO Director shall publish guidance to patent examiners on use of AI in the inventive process and issue additional guidance to patent examiners on other AI issues.
- Secretary of Homeland Security shall develop: (1) training, analysis, and an evaluation program to mitigate AI-related IP risks and (2) a plan for multilateral engagements to encourage the adoption of the AI safety and security guidelines for use by critical infrastructure owners and operators.
- Secretary of Labor shall develop and publish best practices for employers to mitigate AI’s potential harms to employees’ well-being.
- Attorney General shall recommend best practices for law enforcement, including safeguards and appropriate limits on AI use.
- Interagency group created by Section 3 of this order shall identify and share best practices for recruiting and hiring law enforcement professionals with technical skills and for training law enforcement officials.
- Secretary of Health and Human Services shall direct HHS to: (1) develop a strategy to determine whether AI-enabled technology in the health and human services sector maintains appropriate levels of quality, and (2) the Secretary of HHS shall develop a strategy for regulating use of AI in drug development.
- Administrator of General Services shall develop and issue a framework for prioritizing critical and emerging technologies in the Federal Risk and Authorization Management Program authorization process, starting with generative AI and is encouraged to prioritize generative AI.
- Secretary of Education shall develop resources and policies regarding AI that includes an AI toolkit for implementing recommendations from the Department of Education’s AI and the Future of Teaching and Learning report.
- AI and Technology Talent Task Force shall identify and circulate best practices for agencies to attract, hire, retain, train, and empower AI talent and shall plan and budget for AI needs.
- Director of OPM shall develop guidance on the use of generative AI for work by federal workforce.
- Directors of OPM and OMB shall: (1) issue guidance for agency application of existing pay flexibilities or incentive pay programs to attract AI talent and (2) establish guidance and policy on skills-based, government-wide hiring of AI talent.
The order requires several agencies to either publicly release or submit to the President reports about best practices for AI use, effects of AI, potential for increasing AI use within the federal government, and recommendations for further executive action on AI.
Reports to the public
- Treasury shall issue public report on best practices for financial institutions to manage AI risk.
- Secretary of Energy shall issue a public report describing AI potential to improve electric grid infrastructure and to provide clean electric power to all Americans.
- President’s Council of Advisors on Science and Technology shall make publicly available (and submit to the President) a report on the potential role of AI in research aimed at tackling major societal/global challenges.
Reports to the President/Executive Branch
- President’s Council of Advisors on Science and Technology shall submit to the President (and make public) a report on the potential role of AI in research aimed at tackling major societal/global challenges.
- Council of Economic Advisers chairman shall submit report to the President on labor-market effects of AI.
- Secretary of Labor shall submit to the President a report analyzing agencies’ abilities to support workers displaced by AI adoption.
- Attorney General shall submit to the President a report that addresses use of AI in the criminal justice system.
- Secretary of Commerce shall submit: (1) a report to OMB Director and Assistant to the President for National Security Affairs identifying the existing tools and potential development of techniques for authenticating content, tracking its provenance, watermarking, detecting synthetic content, preventing generative AI from producing child sexual abuse material or nonconsensual intimate images, testing software, and auditing synthetic content; (2) a report to the President on the potential benefits, risks, and implications of dual-use foundation models for which the model weights are widely available, as well as policy and regulatory recommendations pertaining to those models, incorporating public input; and (3) a report to President on priority actions taken to engage globally and develop standards on AI.
- Defense secretary shall submit report to the President through the Assistant to the President for National Security Affairs that includes recommendations to address challenges to the Department of Defense’s ability to hire noncitizens, recommendations to streamline processes for accessing classified info for certain noncitizens, recommendations for appropriate use of enlistment authority for AI experts, and recommendations for the Departments of Defense and Homeland Security to work together to enhance use of authorities for retention of certain noncitizens of vital importance to national security.
- AI and Technology Talent Task Force shall track and report progress to the President on increasing AI capacity across federal government.
- Secretary of Homeland Security shall submit a report to the President on priority actions to mitigate cross-border risks to critical U.S. infrastructure.
- Under Secretary of Commerce for Intellectual Property and USPTO Director shall consult with Director of U.S. Copyright Office and issue recommendations to the President on potential executive orders on copyright and AI.
The Executive Order established the White House Artificial Intelligence Council, which will coordinate activities of agencies across the federal government, and named the Assistant to the President and Deputy Chief of Staff for Policy as the Chair.
The order also tasks agencies with forming new AI-focused committees.
- Secretary of Homeland Security shall establish Artificial Intelligence Safety and Security Board.
- Secretary of Health and Human Services shall establish an HHS AI Task Force to develop a strategic plan, including policies and frameworks on responsible deployment and use of AI in the health sector, and identify resources to promote deployment.
- OMB Director shall convene and chair an interagency council to coordinate development and use of AI in agencies’ programs and operations.
- Assistant to the President and Deputy Chief of Staff for Policy shall convene an AI and Technology Talent Task Force to accelerate and track the hiring of AI and AI talent across federal government.
- Secretary of Transportation shall direct Nontraditional and Emerging Transportation Technology Council to establish a new DOT Cross-Modal Executive Working Group.
- Directors of OPM and OMB shall establish an interagency working group to facilitate government-wide hiring of people with AI and technology skills.
Tangible steps for agencies
While a significant portion of the Executive Order is focused on the development of future guidance, the order includes a number of tangible steps that agencies are tasked with completing as well. Key examples include launching several pilot programs, streamlining agency processes, and prioritizing funding for AI.
- All agencies must: (1) designate a Chief Artificial Intelligence Officer; (2) create internal Artificial Intelligence Governance Boards; (3) implement or increase AI training that gives employees operating knowledge of AI; and (4) conduct a security review of all data assets in the comprehensive data inventory and take steps to address highest-priority potential security risks that releasing that data could raise.
- Director of NSF shall: (1) launch a pilot program of NAIRR to make resources available to the research community; (2) identify data appropriate for inclusion in the pilot; (3) shall fund and launch at least one NSF Regional Innovation Engine to prioritize AI work; (4) establish at least four new National AI Research Institutes; (5) fund creation of a Research Coordination Network dedicated to PETs; (6) engage with agencies to identify opportunities to incorporate PETs into operations and prioritize research that encourages the adoption of leading-edge PETs; and (7) prioritize available resources to support AI education and workforce development and consult with agencies to identify further opportunities for agencies to allocate resources.
- Heads of agencies that the Director of NSF identifies as having data appropriate for inclusion in the pilot program shall submit a report identifying agency resources that could be integrated into the pilot.
- Commerce secretary shall: (1) require companies developing potential dual-use foundation models to give the government information on training, development, and production of models, protections taken to assure integrity of training, ownership and possession of model weights of dual-use foundation models and measures taken to protect them, and results of red-team testing; (2) require anyone who acquires, develops, or produces a potential large-scale computing cluster must report that and its total computer power; (3) lead preparations for a coordinated effort with key international allies and standards development organizations to drive development and implementation of AI-related consensus standards and info-sharing; (4) establish a plan for global engagement on promoting and developing AI standards; and (5) implement a flexible membership structure for National Semiconductor Technology Center, implement mentorship programs to increase participation in the industry, increase availability of resources to small businesses and startups, and consider the inclusion of competition-increasing measures in notices of funding availability for commercial R&D facilities focused on semiconductors to promote competition in semiconductor industry.
- OMB director shall: (1) identify commercially available information procured by agencies (including from data brokers) in agency reporting processes; (2) evaluate agency standards and procedures associated with the collection, processing, maintenance, use, sharing, dissemination, and disposition of commercially available info containing PII to inform potential guidance to agencies; (3) take steps to support short- and long-term strategy identified through the RFI; (4) issue guidance to agencies for labeling and authenticating content they produce or publish to strengthen public confidence in US government content; and (5) issue instructions to agencies for collecting, reporting, publishing agency AI use cases.
- Secretaries of Defense and Homeland Security shall develop plans for, conduct, and complete an operational pilot project to identify, develop, test, evaluation, and deploy AI to aid in discovery and remediation of vulnerabilities in government tech.
- Secretaries of State and Homeland Security shall streamline processing times of visa applications and facilitate continued availability of visa appointments for applicants with AI expertise.
- Secretary of Homeland Security shall: (1) review and initiate any policy changes necessary to modernize immigration pathways for AI experts and (2) develop and publish informational resources to attract and retain AI experts that will be published on ai.gov.
- Secretary of Health and Human Services shall: (1) identify and prioritize grantmaking to support responsible AI development and use; (2) publish a plan to address automated decision-making in public benefits to promote assessment of access to benefits, notice to recipients about use of automated decision-making, regular evaluation to detect unjust denials, processes to retain appropriate levels of discretion, processes to appeal denials to humans, and analysis of whether automated decision-making in public benefits is equitable; (3) establish an AI safety program that establishes a common framework to approaches to identify and capture clinical errors from AI deployed in health care, analyzes captured data to develop best practices to avoid harms, and disseminate those recommendations to appropriate stakeholders.
- Secretary of Veterans Affairs shall host two 3-month nationwide AI Tech Sprint competitions.
- Secretary of Energy shall: (1) collaborate with the private sector to support development of AI tools to mitigate climate change risks; (2) expand partnerships to utilize Energy’s computing capabilities and AI testbeds to build foundation models that support new applications in science and energy and national security; (3) establish an office to coordinate development of AI across Energy programs and the 17 National Laboratories; and (4) establish a pilot program to enhance existing successful training programs for scientists.
- Small Business Administration administrator shall: (1) prioritize allocation of Regional Innovation Cluster program funding for establishing one or more Small Business AI Innovation and Commercialization Institutes; (2) prioritize allocation of up to $2M in Growth Accelerator Fund Competition bonus prize funds to support of AI-related curricula/training; (3) assess the extent to which eligibility criteria of existing programs support appropriate expenses by small businesses related to adoption of AI; and (4) conduct outreach regarding opportunities for small businesses to use capital-access programs for eligible AI purposes.
- Secretary of Labor shall: (1) issue guidance to make clear to employers using AI that workers are still to be compensated for their hours worked and (2) publish guidance for federal contractors regarding nondiscrimination in hiring involving AI.
- Attorney General shall review Executive Order 14074 and reassess existing agency capacity to investigate law enforcement deprivation of rights resulting from AI use.
- Secretary of Agriculture shall issue guidance to public benefits administrators on the use of automated decision-making in implementing benefits or providing customer service.
- Secretary of Housing and Urban Development shall (and CFPB is encouraged to): (1) issue additional guidance on the use of tenant screening systems in ways that may lead to discriminatory outcomes that violate federal law and (2) address how the Fair Housing Act, Consumer Financial Protection Act, and Equal Credit Opportunity Act apply to online ads that use AI.
- Head of agencies with authority over critical infrastructure and CISA shall evaluate and provide the Department of Homeland Security an assessment of potential risks related to AI in critical infrastructure and consider mitigations.
- Assistant to President for National Security Affairs and Assistant to President and Deputy Chief of Staff for Policy shall oversee interagency process for developing and submitting a proposed National Security Memo on AI to the President that outline actions for relevant agencies to address national security risks and benefits of AI.
- OSTP and OMB directors shall identify priority mission areas for increased federal government AI talent.
- Tech talent programs shall develop and begin to implement plans to support rapid recruitment of government-wide AI talent surge to accelerate placement of key AI talent and to advance agencies’ tech strategies.
- Chief Data Officer Council shall develop a position-description library for data scientists and a hiring guide.
- AI and Technology Talent Task Force shall: (1) coordinate the use of fellowship programs and agency tech talent programs to place AI talent to fill staffing gaps and (2) convene a cross-agency forum for ongoing collaboration between AI professionals to share best practices and improve retention.
- Secretary of State shall: (1) establish a program to identify and attract top AI talent; (2) lead efforts outside of military and intelligence to expand engagements with international allies to advance those allies’ understanding of US AI policies; and (3) lead efforts to establish a strong international framework for managing risks and benefits of AI by encouraging voluntary commitments.
- Secretary of State and Administrator of the U.S. Agency for International Development shall: (1) publish an AI in Global Development Playbook that incorporates the AI RMF and (2) develop a Global AI Research Agenda.
- Secretary of Transportation shall: (1) direct Nontraditional and Emerging Transportation Technology Council to assess the need for information regarding the use of AI in transportation, support initiatives to pilot transportation-related applications of AI, and evaluate the outcomes of these pilot programs to recommend appropriate regulatory action; (2) direct DOT advisory committees to produce advice on safe and responsible use of AI in transportation; and (3) direct ARPA-I to explore transportation-related opportunities and challenges of AI and encourage ARPA-I to prioritize allocation of grants to those opportunities.
- Directors of OPM and OMB shall: (1) conduct evidence-based review on need for hiring and workplace flexibility; (2) coordinate a pooled-hiring action to support AI talent recruitment across agencies; (3) review existing Executive Core Qualifications for Senior Executive Services positions informed by AI literacy competency and implement new ECQs as appropriate; and (4) complete a review of competencies for civil engineers and make recommendations for ensuring that adequate AI expertise in these occupations in government reflects increased use of AI in critical infrastructure.
Many of the specific details about the implementation of this EO will be laid out in the OMB guidance. EPIC will publish an analysis of the OMB draft guidance soon.