CSET

Ahead of this week's Trump-Xi summit, @jacob_feldgoise shares his thoughts with the @nytimes on China's latest AI chip advances and what they mean for the U.S. nytimes.com/2026/05/12/bus…

📢CSET is hiring!📢 Don't miss your chance to join our fantastic team and work at the intersection of tech and national security. We are looking for a UX Engineer, a Data Scientist, and a Fellow to lead our work on Frontier AI. Details below: cset.georgetown.edu/careers/

Lawrence has spearheaded leadership of CSET’s Inclusion Alliance, which is committed to fostering a workplace where all CSETers feel valued, respected, and supported. Congrats again, Lawrence, on this well-deserved award!

Congrats to CSET's own Lawrence Hailes for winning this year's @georgetownsfs Community in Diversity Individual Leader Award, which recognizes outstanding individuals who have created a more inclusive School of Foreign Service community.

In their piece in @barronsonline, @steph_batalis, Katherine Quinn, and Rebecca Gelles paint a picture of the key role federal research investments play in spurring U.S. innovation competitiveness. Be sure to check out their latest analysis below! barrons.com/articles/gover…

⭐️New Report⭐️ As AI introduces new risks, some potentially catastrophic or even existential, there is little data or detailed theory to assess them. Our report helps to shed light on this challenge and offers helpful solutions for policymakers. cset.georgetown.edu/publication/be…

🎉New job alert!🎉 Come join our team as a Data Scientist and build tools for identifying emerging areas in S&T. If you want to work on cutting-edge data science at the intersection of national security and tech be sure to apply below! cset.georgetown.edu/job/data-scien…

Check out @KathleenCurlee's latest op-ed arguing for an "Artemis Alliance" to support and protect the space-based infrastructure that is key to modern life!

“It is not clear yet who benefits between attackers and defenders," @AndrewJLohn testified to @USCC_GOV in discussing the potential for AI to boost Chinese cyber theft. Read more on this and other emerging cyber threats in his full testimony below. uscc.gov/sites/default/…

What do the PLA’s latest challenges and competitions reveal about emerging technologies such as AI, drones, and cyber? @jules__george’s new CSET blog explores multi-domain integration, unmanned systems, and military-civil fusion. cset.georgetown.edu/article/chinas…

✨New CSET blog ✨ on the PLA’s latest efforts on challenges and competitions on emerging technologies: @jules__george highlights key takeaways on multi-domain integration, unmanned technological innovation and countermeasures, and further evidence of military-civil fusion. cset.georgetown.edu/article/chinas…

CSET is proud to announce that Research Fellow Sam Bresnick has been selected as a Fellow for the 2026 cohort of @PennUSChina. Congratulations, Sam!

What counts as an “AI job”? The answer shapes how we measure talent shortages, design workforce policy, and understand AI’s labor market impact. CSET’s new blog compares prevailing approaches and explains CSET’s new definition of AI development jobs: cset.georgetown.edu/article/defini…

Recently, there has been a lot of interest in distillation, particularly regarding “adversarial” distillation from Chinese labs. At the core of this issue is the question of ‘how much performance gains (ie uplift) do student models achieve when they are distilled from closed-source teacher models.’ Months ago, we started reviewing the literature to try and answer this question. We’ve so far reviewed ~40 papers on black-box distillation, focusing on papers that distill the reasoning, math, and code capabilities of closed-source teacher models. All white-box distillation research is omitted. But there's a problem. The data is not straightforward. Here are some of the challenges we ran into, and why the distillation literature doesn't paint a clear picture (long 🧵): 1/ Old research. To start, much of the literature on this topic – that we reviewed – was published between 2022-2024. The student and teacher models are old, and the benchmarks used to evaluate performance are saturated. And even newer papers often used older models. This means that we can’t simply take the data from the literature and assume it applies to Chinese "adversarial" distillation today. Some student models saw extreme performance gains, while, at times, others saw degraded performance on out-of-distribution benchmarks. In the most extreme case we identified, a distilled model improved by 25600% relative to its baseline score; the student went from 0.2 to 51.2 on the ASDIV benchmark (this is from the Program-Aided Distillation paper). But that large percentage is due to the poor baseline performance it started with. The ASDIV benchmark is saturated, the student was a 0.22B parameter CodeT5 model, and the teacher was GPT-3.5-Turbo. In my view, we cannot infer much from this example because it's not applicable to frontier models and challenging unsaturated benchmarks. We saw this same issue across most of the papers we reviewed. Most of the student models were years old and relatively small (<13B params). Llama-1/2 and CodeLLama were the most common students, and GPT-3.5/4 the most common teachers (likely because OpenAI permits distillation in their ToS for research purposes). On average across the papers, student models reached 74% of the teacher’s performance (primarily related to older code, math, and reasoning benchmarks), with a high standard deviation. But again, we can’t really infer too much from this, and we cannot assume this is the uplift China gets on frontier benchmarks. It's safe to say they see some degree of uplift, but we already knew that (they probably wouldn't distill in the first place if they didn't see any gains from it). 2/ The next issue is that most of the literature is small-scale, experimental, and focused on compression. None of it is “adversarial,” rather the intent is to compress knowledge into smaller parameter models. The difference in intent in the literature vs. Chinese distillation matters. To highlight my earlier point, most of the literature is trying to compress knowledge into small <13B parameter students. That doesn't translate to distilling into, say, the 1 trillion param Kimi-K2.5 model. Moreover, the quantity (and quality) of the synthetic teacher-generated data is likely different from what Chinese labs are using. Anthropic identified distillation campaigns involving over 16 million exchanges with Claude, likely generating billions of tokens. The literature, on the other hand, is much smaller scale. Back-of-the-envelope calculations show that, on average, researchers were likely distilling on ~185 million tokens (these are rough calculations). Only a few papers distilled with over a billion tokens.

✨New blog✨ Who counts as part of the AI workforce? Existing definitions often blur together workers who build AI systems, deploy them, or simply use AI tools. CSET’s new blog explains why that matters, and introduces CSET’s new definition of AI development jobs: cset.georgetown.edu/article/defini…

.@colemcfaul and I have documented the Chinese military's interest in @nvidia chips; it's clear that H200s will contribute to the PLA’s modernization, either through direct purchases or through the use of LLMs trained on them. Link below. @CSETGeorgetown @emergingtechobs

New research from McCourt Associate Research Professor Renée DiResta and @CSETGeorgetown's Josh Goldstein breaks down the rise of full-spectrum propaganda. 🧵

Tune in at 10:15 to watch @hlntnr discuss China's theft of U.S. innovation with @SenJudiciaryGOP and @JudiciaryDems! judiciary.senate.gov/committee-acti…

NEW @CSETGeorgetown + @emergingtechobs piece! Does China's access to US semiconductor technology help the PLA develop and deploy military AI? After 3 years reading thousands of PLA procurement docs, @sambresnick and I say yes. Here’s how, and why it matters: 🧵/13

"Well-designed governance does not suppress innovation. Instead, it shapes the direction of innovation in socially beneficial ways..." @oschinski and @MinaNrn reframe the discussion around AI innovation and governance. Read more in @Newsweek below: newsweek.com/stop-asking-wh…