Key AI Accomplishments in the Year Since the Biden-Harris Administration’s Landmark Executive Order
One year ago, President Biden issued a landmark Executive Order to ensure that America leads the way in seizing the promise and managing the risks of artificial intelligence (AI). The Executive Order directed sweeping actions to manage AI’s safety and security risks, protect Americans’ privacy, advance equity and civil rights, stand up for consumers and workers, promote innovation and competition, advance American leadership around the world, and more.
Today, the Biden-Harris Administration is announcing that Federal agencies have completed on schedule each action that the Executive Order tasked for this past year—more than one hundred in all. Below are some of the Administration’s most significant accomplishments on managing AI’s risks and seizing its promise in the year since President Biden signed his Executive Order.
Managing Risks to Safety and Security:
The Executive Order directed the boldest actions ever taken to protect Americans from a broad range of AI’s safety and security risks, including risks related to dangerous biological materials, software vulnerabilities, and foreign actors’ efforts to develop AI for harmful purposes. Over the last year, to protect safety and security, agencies have:
Used Defense Production Act authorities to require developers of the most powerful AI systems to report vital information, including results of safety and security testing, to the U.S. government. These companies have notified the Department of Commerce about the results of their red-team safety tests, their plans to train powerful models, and large computing clusters they possess capable of such training. Last month, the Department of Commerce proposed a rule to require the reporting of this information on a quarterly basis.
Led the way on AI safety testing and evaluations to advance the science of AI safety. The U.S. AI Safety Institute (US AISI) at the Department of Commerce has begun pre-deployment testing of major new AI models through recently signed agreements with two leading AI developers. The Department of Energy (DOE) developed and expanded its AI testbeds and evaluation tools, which it has already used to test models’ risk to nuclear security.
Developed guidance and tools for managing AI risk. The US AISI and the National Institute of Standards and Technology (NIST) at the Department of Commerce published frameworks for managing risks related to generative AI and dual-use foundation models, and earlier this month, AISI released a Request for Information on the responsible development and use of AI models for chemical and biological sciences. The Department of Defense (DoD) released its Responsible AI toolkit to align AI projects with the Department’s Ethical Principles.
Issued a first-ever National Security Memorandum (NSM) on AI. The NSM directs concrete steps by Federal agencies to ensure the United States leads the world’s development of safe, secure, and trustworthy AI; to enable agencies to harness cutting-edge AI for national security objectives, including by protecting human rights and democratic values; and to advance international consensus and governance on AI. This essential document serves as a formal charter for the AI Safety Institute, designating it as the center of the whole-of-government approach to advanced AI model testing, and will guide rapid and responsible AI adoption by the DoD and Intelligence Community. The NSM also directs the creation of a Framework to Advance AI Governance and Risk Management in National Security, which provides agile guidance to implement the NSM in accordance with democratic values, including mechanisms for risk management, evaluations, accountability, and transparency.
Finalized a framework for nucleic acid synthesis screening to help prevent the misuse of AI for engineering dangerous biological materials. The framework, developed by the Office of Science and Technology Policy (OSTP), encourages nucleic acid synthesis providers to identify gene sequences that could be used to pose national security risks, and to implement customer screening to mitigate the risks of misuse. Federal agencies will require that funding recipients obtain synthetic nucleic acids from vendors that adhere to the framework, starting in 2025. The Department of Homeland Security (DHS) has developed an initial framework with principles for evaluating the effectiveness of screening mechanisms going forward.
Launched a new Task Force on AI Datacenter Infrastructure. The Task Force provides streamlined coordination on policies to advance datacenter development operations in line with economic, national security, and environmental goals.
Identified measures—including approaches for labeling content and improving transparency—to reduce the risks posed by AI-generated content. The Department of Commerce submitted to the White House a final report on science-backed standards and techniques for addressing these risks, while NIST has launched a challenge to develop methods for detecting AI-generated content. President Biden has emphasized that the public has a right to know when content is AI-generated, and agencies are working to use these tools to help Americans to know that communications they receive from their government are authentic.
Combatted AI-generated image-based sexual abuse. Image-based sexual abuse—both non-consensual intimate images of adults and child sexual abuse material—is one of the fastest growing harmful uses of AI to date and disproportionately targets women, children, and LGBTQI+ people. This year, following the Vice President’s leadership in underscoring the urgent need to address deepfake image-based sexual abuse and a White House Call to Action to reduce these risks, leading AI developers and data providers made voluntary commitments to curb the creation of AI-generated image-based sexual abuse material. Additionally, the Department of Justice (DOJ) funded the first-ever helpline to provide 24/7 support and specialized services for victims of the non-consensual distribution of intimate images, including deepfakes. The Department of Education also clarified that school responsibilities under Title IX may extend to conduct that takes place online, including AI-generated abuse.
Established the AI Safety and Security Board (AISSB) to advise the Secretary of Homeland Security on the safe and secure use of AI in critical infrastructure. The AISSB has met thrice this year to develop a set of recommendations for entities that develop, deploy, and promote accountability for AI systems that assist in delivering essential services to millions of Americans. The work of the AISSB complements DHS’s first-ever AI safety and security guidelines for critical infrastructure owners and operators, which were informed by agencies’ assessments of AI risks across all critical infrastructure sectors. To help protect critical infrastructure further, the Department of Treasury released a report on managing security risks of AI use in the financial sector, and the Department of Energy released an assessment of potential risks to the power grid, as well as ways in which AI could potentially strengthen grid resilience and our ability to respond to threats.
Piloted AI for protecting vital government software systems. The Department of Defense and DHS conducted AI pilots to address vulnerabilities in government networks used, respectively, for national security purposes and for civilian governmental organizations.