Photo: Pixabay
The White House has issued a memo urging U.S. intelligence agencies to harness the power of artificial intelligence while managing its risks.
THE note is the first in the federal government to address national security specifically in relation to AI. This follows President Joe Biden’s executive order on AI in 2023.
The memo focuses on three areas: ensuring the United States is a global leader in developing safe, secure, and trustworthy AI; using cutting-edge AI to advance U.S. national security interests; and build international consensus on AI governance.
“AI has become a revolutionary technology and has demonstrated significant and growing importance to national security,” the memo said. “AI, if used appropriately and for its intended purposes, can provide great benefits.
“If misused, AI could threaten U.S. national security, reinforce authoritarianism around the world, undermine democratic institutions and processes, facilitate human rights abuses, and weaken order rules-based international framework. »
Learn more: OECD launches G7 toolkit for ‘safe, secure and trustworthy’ AI in the public sector
And then?
The memo outlines a series of actions for agencies to accelerate the development of AI.
These include a call for the National Science Foundation to use the National AI Research Resource to “distribute computing resources, data, and other assets critical to the development of AI” to universities, non-profit organizations and independent researchers who might not otherwise have access to it.
It also directs the Department of Education to “launch a pilot project to evaluate the performance and effectiveness of federated AI and data sources for training, fine-tuning, and inference at scale.” ‘AI’.
In the interest of protecting U.S.-based AI systems from hostile threats abroad, the memo says it is essential that the federal government “ensure the fundamental capabilities across the United States that power development of AI” and “provide appropriate government assistance to relevant non-governmental organizations”. government entities.”
To achieve this, members of the intelligence community, including the Office of the Director of National Intelligence and the Departments of Defense, Justice, Commerce, Education and Homeland Security, are asked to “identify critical nodes of the AI supply chain and to develop a list. of the most plausible pathways by which these nodes could be disrupted or compromised by foreign actors.”
The memo also warns that foreign actors pose a risk to U.S. intellectual property through what it calls “gray zone methods,” such as “technology transfer and data localization requirements.”
“AI-related intellectual property often includes critical technical artifacts (CTAs) that would significantly reduce the costs of recreating, obtaining, or using powerful AI capabilities,” the memo states. “The United States government must guard against these risks. »
Learn more: France appoints its first AI minister
Foundations laid
Earlier this year, the US federal cybersecurity agency named its first director of AI, Lisa Einsteinto demonstrate its “commitment to using AI responsibly to advance its cyber defense mission.”
Einstein had previously served as a senior advisor to the Cybersecurity and Infrastructure Security Agency and executive director of its Cybersecurity Advisory Committee. She was responsible for producing and implementing an AI roadmap promoting the use of AI in cybersecurity work.
Earlier this month, Office of Management and Budget issued new procurement guidelines for agencies to improve risk management in purchasing AI systems and machine learning software.
These include recommendations for agencies to assess the risk any acquisition poses in terms of “privacy, security (and) data ownership and rights”.
In September, the the first global treaty on AI was signed by national governmentsnotably in the US, UK and EU member states, with a focus on protecting and integrating human rights into existing and new AI technologies.
Although hailed as the first legally binding global treaty on AI, the AI Convention has also been criticized by activists for granting exemptions to AI used in the interests of national security.
