On August 9, 2023 the Biden administration announced an executive order to restrict certain U.S. investments in China’s key technology sectors, including artificial intelligence. This blog post proposes implementing investment restrictions related to AI for national security using a list-based end-user approach that builds upon existing list-based tools.

The Summer of 2023 will likely go down in the history books as the “Summer of AI”—with numerous advancements in LLMs and other generative AIs capturing the public’s consciousness. At the same time, the Biden administration announced multiple efforts to ensure the American public’s safety, security, and privacy, while ensuring the U.S. continues to lead the world in AI innovation. This blog post summarizes some of these major executive branch actions and highlights related CSET insights.

As policymakers decide how best to regulate AI, they first need to grasp the different types of harm that various AI applications might cause at the individual, national, and even societal levels. To better understand AI harm, this blog post presents some key components and characteristics.

The much-anticipated National Cyber Workforce and Education Strategy (NCWES) provides a comprehensive set of strategic objectives for training and producing more cyber talent by prioritizing and encouraging the development of more localized cyber ecosystems that serve the needs of a variety of communities rather than trying to prescribe a blanket policy. This is a much-needed and reinvigorated approach that understands the unavoidable inequities in both cyber education and workforce development, but provides strategies for mitigating them. In this blog post, we highlight key elements that could be easily overlooked.

Drawing from prior CSET research, this blog post describes different domestic and international initiatives the Chinese government and companies are pursuing to shore up investment in AI and meet China’s strategic objectives, as well as indicators to track their future trajectories.

Large language models (LLMs) could potentially be used by malicious actors to generate disinformation at scale. But how likely is this risk, and what types of economic incentives do propagandists actually face to turn to LLMs? New analysis uploaded to arXiv and summarized here suggests that it is all but certain that a well-run human-machine team that utilized existing LLMs (even open-source ones that are not cutting edge) would save a propagandist money on content generation relative to a human-only operation.

How can we measure the reliability of machine learning systems? And do these measures really help us predict real world performance? A recent study by the Stanford Intelligent Systems Laboratory, supported by CSET funding, provides new evidence that models may perform well on certain reliability metrics while still being unreliable in other ways. This blog post summarizes the study’s results, which suggest that policymakers and regulators should not think of “reliability” or “robustness” as a single, easy-to-measure property of an AI system. Instead, AI reliability requirements will need to consider which facets of reliability matter most for any given use case, and how those facets can be evaluated.

CSET has received a lot of questions about LLMs and their implications. But questions and discussions tend to miss some basics about LLMs and how they work. In this blog post, we ask CSET’s NLP Engineer, James Dunham, to help us explain LLMs in plain English.

Collaborations between researchers and policymakers are necessary for progress, but can be challenging in practice. This blog post reports on recent discussions by privacy experts on the obstacles they face when engaging in the policy space and advice on how to improve these barriers.

On July 21, the White House announced voluntary commitments from seven AI firms to ensure safe, secure, and transparent AI. CSET’s research provides important context to this discussion.