The future of artificial intelligence holds both extraordinary promise and peril. With the stroke of a pen, President Biden set forth a monumental executive order that aims to steer the nation towards realizing AI's benefits while mitigating its risks. This bold action reflects the urgency of governing AI development responsibly as capabilities rapidly advance. The order lays the foundation for a coordinated effort across government, academia, private industry and civil society to harness AI for good. It is a call to action for America to lead in shaping AI to reflect our ideals of justice, security and opportunity. The principles outlined in the executive order make clear: AI is only as just, creative and decent as the people who build it and the data on which it is trained. This moment demands the best of America's ingenuity, diversity and moral character to ensure AI promotes, not undermines, our democratic values and aspirations. The choices made today will indelibly shape the society of tomorrow. With this executive order, President Biden lights the path forward to an AI-powered future that empowers people, energizes the economy, and enhances our collective wellbeing. The destination is clear, the stakes are high; now begins the work to get there.
Section 1:
The order establishes 8 guiding principles for responsible AI development and use: safety/security, innovation/collaboration, supporting workers, equity/civil rights, consumer protection, privacy/civil liberties, government capacity, and global leadership.
It calls for robust testing and evaluations of AI systems before deployment to ensure they are secure, ethical and compliant with laws.
The order aims to promote investments in AI research, education and training to maintain US leadership while providing opportunities for all Americans.
It emphasizes using AI to advance equity and civil rights, not exacerbate discrimination. Accountability measures for developers/deployers are needed.
Protecting privacy and civil liberties is declared a priority, including through use of privacy-enhancing technologies.
Building AI expertise and capacity within government is deemed critical for proper regulation and governance.
The order seeks to spearhead global coordination on AI safety, ethics and governance to ensure responsible development that benefits humanity.
Overall the order lays out a comprehensive vision and principles for steering AI innovation to reflect democratic values and aspirations. It signals strong federal leadership is crucial to realizing AI's promise while mitigating risks.
Section 2:
It establishes 8 guiding principles for responsible AI: safety/security, innovation/competition, supporting workers, equity/civil rights, consumer protection, privacy/civil liberties, government capacity, and global leadership.
AI systems should be rigorously tested and evaluated before deployment to ensure they are secure, ethical and compliant with laws.
Investments should be made in AI research, education and training to maintain US leadership while providing opportunities for all Americans.
AI must not exacerbate discrimination or inequality; accountability measures are needed.
Protecting privacy and civil liberties is a priority, including through privacy-enhancing technologies.
Building expertise and capacity within government is critical for proper AI governance.
The US should lead globally on AI safety, ethics and governance standards.
The order provides key definitions for terms like AI, AI model, AI system, dual-use AI, generative AI, machine learning, synthetic content, and more.
Overall it lays out a comprehensive vision and principles for steering AI innovation to reflect democratic values, with government leadership key to realizing the promise while mitigating risks.
Section 3:
It directs the development of guidelines and standards for safe, secure and trustworthy AI systems. This includes guidance on AI security, testing and auditing.
Reporting requirements are established for companies developing potential dual-use AI models and large computing clusters. This aims to ensure models are secure and not misused.
Rules are proposed to increase oversight of foreign use of US cloud computing services to train AI models, to prevent malicious uses.
AI's risks and benefits for critical infrastructure security are to be evaluated. Guidance will be updated to incorporate AI security best practices.
Pilot projects are initiated to test using AI to find and fix software vulnerabilities in government systems.
Reviews are ordered on biosecurity risks from AI and synthetic biology. Steps are mandated to strengthen screening for concerning synthetic DNA orders.
Agencies must review risks of releasing open government data that could aid development of weapons when combined with other data.
A National Security Memorandum is to be developed on governing national security uses of AI, outlining safety, ethics and risk management principles.
It aims to balance AI innovation and open data access with new safeguards, risk assessments and oversight to prevent misuse and malicious applications.
The order establishes reporting requirements for companies training large AI models that could have dual use capabilities that pose security risks, such as developing biological weapons or cyberattacks.
It directs actions to prevent generative AI from producing harmful synthetic content like child sexual abuse material or non-consensual intimate imagery.
Guidance is ordered on authenticating digital content and labeling synthetic media to strengthen public confidence.
It calls for evaluating the risks and benefits of dual-use foundation models that have publicly available model weights, which could aid innovation but also remove safeguards.
The order aims to build AI expertise and capacity within government agencies to properly regulate and govern AI technologies.
It frames US leadership in AI governance as critical and directs coordination with allies to develop shared principles and approaches for managing risks.
Overall, the order takes a comprehensive, proactive approach to balancing AI innovation with ethics, safety and security across multiple domains.
Section 4:
It directs steps to attract and retain AI talent in the US, including streamlining visa processes and establishing new visa programs.
New institutes and training programs are launched to strengthen US AI research and develop an AI-ready workforce.
Guidance is ordered on AI's implications for inventorship and patent eligibility to promote innovation.
Programs are established to help small businesses and entrepreneurs adopt AI tools and protect AI-related IP.
Initiatives are launched across sectors - healthcare, energy, climate, veterans affairs - to responsibly advance AI applications.
Competition policy is directed at semiconductor manufacturing, AI marketplaces, and supporting small business adoption of AI.
Overall, it aims to strengthen the US AI ecosystem by removing barriers, expanding access and opportunities, and steering innovation to reflect American values.
A key theme is enhancing inclusion - of workers, small businesses, and underserved communities - in the AI revolution.
It portrays US leadership in AI as requiring both technical prowess and responsible governance to earn public trust.
Section 5:
Understanding AI’s Implications on Labor:
Chairman of the Council of Economic Advisers to draft a report on AI's labor market effects.
Secretary of Labor to investigate government strategies for AI-caused workforce disruptions.
Existing Programs and AI-related Disruptions:
Evaluate current Federal programs like unemployment insurance and Workforce Innovation and Opportunity Act for AI-related disruptions.
Explore options, including legislation, to:
Increase support for AI-affected workers.
Enhance AI-related education and training opportunities.
Promoting Employees’ Well-being in AI-Integrated Workplaces:
Secretary of Labor to create guidelines for AI in workplaces, covering:
Job displacement risks and AI-related career opportunities.
Maintenance of labor standards, job quality, health, safety, and compensation in AI contexts.
Workers' data rights concerning AI, including transparency and data management.
Agency heads to consider endorsing these guidelines in their programs.
Employers using AI for employee monitoring or augmentation must adhere to compensation laws, especially the Fair Labor Standards Act.
Fostering a Diverse AI-Ready Workforce:
NSF Director to prioritize AI-related education and workforce development.
Collaborate with agencies to identify additional resource allocation opportunities for AI readiness.
Utilize fellowship programs and awards for these purposes.
Section 6:
Advancing Equity and Civil Rights:
Strengthening AI and Civil Rights in the Criminal Justice System:
Attorney General to coordinate with agencies to enforce laws addressing AI-related civil rights violations.
Convene a meeting of Federal civil rights office heads to address discrimination from automated systems and improve public awareness of potential AI biases.
Provide guidance to local investigators on best practices for investigating AI-related civil rights violations.
Attorney General to submit a report detailing the use of AI in various aspects of the criminal justice system, such as sentencing, parole, surveillance, and forensic analysis.
Identify areas where AI can improve law enforcement and recommend best practices to safeguard civil rights.
Promote the recruitment and training of law enforcement professionals with AI expertise.
Protecting Civil Rights Related to Government Benefits and Programs:
Agencies to use civil rights offices to address AI-related discrimination in government programs.
Secretary of HHS to publish a plan addressing the use of automated systems in public benefits administration.
Secretary of Agriculture to issue guidance on the use of automated systems in public benefits programs to ensure equitable outcomes.
Strengthening AI and Civil Rights in the Broader Economy:
Secretary of Labor to publish guidance on non-discrimination in AI-based hiring for Federal contractors.
Federal Housing Finance Agency and Consumer Financial Protection Bureau to evaluate underwriting models and appraisal processes for biases.
Secretary of Housing and Urban Development to issue guidance on tenant screening systems and how they can lead to discriminatory outcomes.
Address how Federal laws apply to advertising on digital platforms using algorithms.
Encourage the Architectural and Transportation Barriers Compliance Board to provide guidance on the use of biometric data in AI for people with disabilities.
Section 7:
Protecting Consumers, Patients, Passengers, and Students:
General Consumer Protection:
Encourage independent regulatory agencies to protect consumers from AI-related risks including fraud, discrimination, and threats to privacy.
Healthcare, Public Health, and Human Services:
Establishment of an HHS AI Task Force to develop policies on AI deployment in health and human services.
Development of AI assurance policy for evaluating AI-enabled healthcare tools.
Promote understanding and compliance with Federal non-discrimination laws related to AI in healthcare.
Establish an AI safety program with Patient Safety Organizations to track and address clinical errors from AI.
Develop a strategy for regulating AI use in drug-development processes.
Transportation Sector:
Assessment by the Nontraditional and Emerging Transportation Technology (NETT) Council on AI in transportation.
Engage Federal Advisory Committees for advice on AI use in transportation.
Direct the Advanced Research Projects Agency-Infrastructure (ARPA-I) to explore AI's transportation-related opportunities and challenges.
Education Sector:
Development of resources, policies, and guidance on safe and non-discriminatory AI uses in education.
Creation of an "AI toolkit" for education leaders.
Communications Networks and Consumers:
Encourage the Federal Communications Commission to consider AI's effects on communications networks.
Examine AI's potential for spectrum management and sharing.
Support efforts to enhance network security using AI.
Encourage efforts to combat unwanted robocalls and robotexts facilitated by AI.
Section 8:
Protecting Privacy:
Director of OMB's Responsibilities:
Evaluate and identify commercially available information (CAI) procured by agencies, especially those containing personally identifiable information.
Consult with the Federal Privacy Council and Interagency Council on Statistical Policy to assess agency standards related to CAI.
Issue an RFI within 180 days to inform potential revisions to guidance on the privacy provisions of the E-Government Act of 2002, particularly concerning privacy impact assessments.
Take necessary steps, based on the RFI process, to issue new or updated guidance or consult with other agencies.
Secretary of Commerce via Director of NIST:
Create guidelines within 365 days for agencies to evaluate the effectiveness of differential-privacy-guarantee protections. These guidelines should address the significant factors and common risks associated with differential privacy.
Director of NSF's Responsibilities:
Fund the creation of a Research Coordination Network (RCN) within 120 days to advance privacy research and promote the development of Privacy Enhancing Technologies (PETs).
Engage with agencies within 240 days to identify ongoing work and opportunities related to PETs and prioritize research for the adoption of leading-edge PETs solutions.
Use the results of the US-UK PETs Prize Challenge to inform approaches for PETs research and adoption.
Section 9:
Advancing Federal Government Use of AI:
Providing Guidance for AI Management
Create an interagency council to coordinate AI development and use across the Federal Government, excluding national security systems.
Within 150 days, issue guidance to strengthen and manage AI use in the Federal Government, which includes:
Designation of a Chief Artificial Intelligence Officer at each agency.
Creation of AI Governance Boards at agencies.
Establishing minimum risk-management practices for AI use.
Recommendations to reduce barriers to AI use.
Testing for generative AI.
Public reporting on compliance.
Develop a method to track agencies' AI progress.
Develop guidelines and tools to support AI risk management and ensure AI systems in contracts align with established guidance.
Improve transparency of agencies’ AI use.
Promote responsible and secure use of generative AI.
Encourage AI investments and facilitate agency access to commercial AI capabilities.
Special provisions for AI in national security systems.
Increasing AI Talent in Government:
Identify priority areas for AI talent recruitment in the Federal Government.
Convene an AI and Technology Talent Task Force to accelerate AI hiring.
Support the rapid recruitment of AI talent across the Federal Government.
Improve Federal hiring practices for AI talent by:
Conducting an evidence-based review on hiring flexibility.
Accelerating the hiring of AI talent.
Establishing guidance on skills-based hiring.
Streamlining personnel-vetting requirements.
Agencies should utilize special authorities to rapidly hire AI talent.
Develop a position-description library for data scientists.
Implement AI training programs for Federal employees.
Address AI talent gaps in national defense, including hiring noncitizens.
Section 10:
Promotion of U.S. AI Policies and Collaboration: The Secretary of State is tasked with:
Expanding engagements with international allies and partners to share U.S. AI-related guidance and policies.
Leading efforts to establish a robust international framework to manage the risks and benefits of AI. This includes encouraging international allies to support commitments similar to those made by U.S. companies.
Global Technical Standards for AI:
The Secretary of Commerce is to prepare a coordinated effort with key international allies and partners for the development of global AI standards.
Within 270 days, a plan for global engagement on AI standards should be established. This plan could cover AI terminology, best practices for data handling, trustworthiness of AI systems, and AI risk management.
A report on priority actions taken according to the plan should be submitted within 180 days of its establishment.
AI in Global Development:
The Secretary of State and the Administrator of the U.S. Agency for International Development (USAID) should create an "AI in Global Development Playbook." This playbook should incorporate AI risk management principles tailored to conditions outside the U.S. borders.
They should also develop a Global AI Research Agenda, guiding AI-related research objectives globally, ensuring responsible AI development and addressing labor-market implications in different contexts.
Addressing AI Risks to Critical Infrastructure:
The Secretary of Homeland Security, in coordination with the Secretary of State, is to lead efforts with international allies to prevent and respond to AI-related disruptions in critical infrastructure systems.
A plan for multilateral engagements to encourage the adoption of AI safety and security guidelines for critical infrastructure should be developed within 270 days.
A report on actions to mitigate cross-border risks to U.S. infrastructure is expected within 180 days of the plan's establishment.
Section 11:
This section introduces the White House Artificial Intelligence Council situated within the Executive Office of the President. Its purpose is to synchronize activities across federal agencies for efficient policy formulation, communication, and AI-related implementations.
Highlights:
Leadership: Chaired by the Assistant to the President and Deputy Chief of Staff for Policy.
Members: Comprises Secretaries from various departments, Directors of agencies such as National Intelligence and NSF, key Presidential Assistants, and other notable roles like the National Cyber Director. The Chair can invite additional members as needed.
Subgroups: The Chair is authorized to form subgroups within the council.
General Provisions:
The order respects other legal authorities and rights and doesn't undermine any executive department or agency's legal authority.
Implementation is subject to available funds and must align with relevant laws.
It doesn't create enforceable rights against the U.S., its entities, or any individual.
Comments