The Center for Security and Emerging Technology within Georgetown University Walsh School of Foreign Service Subscribe: https://nitter.cf/t.co/FjSCGo19xj
Washington, DC
Joined May 2019
- Tweets5.4K
- Following450
- Followers13.3K
- Likes1.6K
Pinned Tweet
⭐️ New Report ⭐️
Red-teaming is a key process in evaluating the safety and security of AI systems. We identify four recurring dimensions following an AI red-teaming exercise and suggest ways policymakers can support the AI test and evaluation ecosystem. cset.georgetown.edu/publicat…
.@OJDaniels told @defense_news that service members need to understand #AI systems’ limitations and recognize when delegating decisions to machines may be inappropriate. defensenews.com/news/pentago…
CSET retweeted
1/ Three states, three frontier AI laws. 🏛️ Using @CSETGeorgetown @emergingtechobs AGORA, @NebhnaniBhumika and I compared California’s SB 53, New York’s RAISE Act, and Illinois’s SB 315 to map how each one governs frontier AI. 🧵
Jessica Ji spoke with @NBCNews about OpenAI agents attempting to change misalignment monitoring settings. “What was interesting to me here is how the instructions degraded and interpretations of them changed as agents delegated to sub-agents — kind of like a game of telephone,” said Ji. nbcnews.com/tech/tech-news/o…
CSET retweeted
How does black-box distillation actually work? In our new blog, we break the distillation process down into five steps based on an in-depth review of the published literature:
1️⃣ Define the objective of the distillation and select the student and teacher models.
2️⃣ Collect seed data that will be inputted into the teacher.
3️⃣ Steer the teacher toward desired outputs.
4️⃣ Elicit synthetic data from the steered teacher by prompting it with seed data.
5️⃣ Train the student on the synthetic data through various techniques.
Since 2022, the U.S. has restricted exports of #AI chips to prevent adversaries from developing advanced AI capabilities. But smugglers have successfully diverted hundreds of thousands of chips.
One proposed solution is location verification. Would it work? Read our report to find out: cset.georgetown.edu/publicat…
Jessica Ji told @FortuneMagazine that autonomous attacks powered by #AI agents are becoming more common, and this is happening because the tools have gotten meaningfully better over the past year or so. “These AI systems are now both more capable of hacking and are able to act more autonomously,” Ji said. fortune.com/2026/10/21/power…
“Agent-led campaigns will probably become more common as the technology continues to improve,” said CSET’s Kyle Crichton to @nytimes for a story on countries applying #AI to their online influence campaigns. nytimes.com/2026/09/18/techn…
Our recent analysis shows that foreign-born scientists and technologists play a substantial role in sustaining America’s scientific and technological leadership. cset.georgetown.edu/publicat…
LAST DAY TO APPLY: Ready to lead a research team focused on the national security applications of #AI? Submit an application today. cset.georgetown.edu/job/nati…
What does it mean to “win” in #AI? A recent report by @AndrewJLohn provides policymakers with a step-by-step process for government leaders to convert AI superiority into actual outcomes that achieve AI’s promise. cset.georgetown.edu/publicat…
What would make publication and patent data more useful and timely for science and technology analysis and decision-making? We hosted a group of experts for an enlightening discussion on this topic. cset.georgetown.edu/publicat…
“The era of unassailable U.S. space power is ending, and Washington has no choice but to adapt,” write @KathleenCurlee and @SamBresnick. Their new op-ed in @ForeignAffairs provides recommendations for U.S. adaptation as China’s space power grows. foreignaffairs.com/china/chi…
CSET retweeted
1/ New piece in @ForeignAffairs with @KathleenCurlee
on China’s rise as a space power and what it means for the United States.
BLUF: The era of unassailable U.S. space dominance is ending, and Washington has no choice but to adapt.
.@hlntnr remarked on barriers to actually implementing an #AI kill switch in @nytimes. nytimes.com/2026/09/19/scien…
.@SamBresnick shares with @BBCWorld his take on China’s attempt to regulate “human-like” #AI chatbots. bbc.com/news/articles/cm4gjy…
CSET’s @OJDaniels spoke with @washingtonpost about President Trump’s new pick for #AI czar, Jay Clayton. washingtonpost.com/wp-intell…
[OCTOBER 21] Join us for “Unlocking S&T Insights,” an event highlighting our Project ATLAS initiative to provide policymakers with the insights they need to keep the U.S. ahead of the curve in technological innovation. Learn more and RSVP: eventbrite.com/e/unlocking-s…
.@SheaBly on @Reuters’ analysis of Chinese-powered #AI agents: “These results provide evidence that the ingredients necessary for an uncontrolled escape are present." reuters.com/business/retail-…
#AI chip smuggling is like an iceberg: we only see what’s above the surface. To estimate its true scale, we need to know how much is hidden below the waterline. Whether that’s 50% or 90% makes a big difference. In this Q&A, @jacob_feldgoise describes how location verification methods can—and cannot—improve enforcement of export controls on these powerful chips. Link below ⬇️
Check out the Q&A: cset.georgetown.edu/article/…
