CSET

⭐️New Report⭐️ How does the critical role of data accessibility factor into unlocking the full potential of military AI? Read more ⬇️ cset.georgetown.edu/publication/be…

Watch the rest of this 2026 @aspenideas Festival panel: cset.georgetown.edu/article/helen-…

CSET Executive Director @hlntnr says the U.S. has a big advantage in #AI diffusion when it comes to enterprise software. #AspenIdeas

Kimi K3 is the largest open AI model ever released. @hlntnr told us why that was always China's plan. Moonshot's July 16 release: 2.8 trillion parameters, a million-token context window, weights public by July 27, and reportedly ahead of Claude Opus 4.8 and GPT-5.5 on the lab's own benchmarks. When we asked the former OpenAI board member who is winning the AI race, she rejected the question itself: 'saying that there's a race actually builds in a bunch of assumptions that there is a finish line.' For most of what matters there is no finish line, and in this clip she maps the competition that does decide things. The frontier-model contest gets the headlines, but the most advanced models are large, slow and expensive to run. The contest underneath is adoption, meaning whose models the world actually builds on. In her words, 'China is really trying to compete in that diffusion and adoption race. And the fact that they're putting out open source models is really helpful for them in getting that adoption.' She also named the labs carrying that strategy on our show, 'obviously famously DeepSeek, but also companies like Moonshot with their Kimi models', before Moonshot became headline news. K3 turns her framing into a live decision for the executives and policymakers in our audience. Once a frontier-class model is free to download, the leaderboard question matters less than the dependency question: whose stack are your teams already building on? More in the first comment. #OpenSource #KimiK3 #USChinaAI @CSETGeorgetown

“The results look very promising,” says CSET Senior Research Analyst Jessica Ji about @OpenAI’s approach to build a red-teaming LLM that helps boost its models’ defenses. #CyberSecurity technologyreview.com/2026/07/15/114…

CSET Executive Director @hlntnr spoke with @washingtonpost’s @ben_guggenheim about the challenges in preventing misuse of open-source #AI models. washingtonpost.com/wp-intelligenc…

CSET Research Fellow @SamBresnick points out why some companies are shifting some workload to cheaper #AI models: “Why would you pay a premium for Anthropic, OpenAI models when for a lot of the workloads you need, the Chinese models are generally workable?” ft.com/content/9c8ff4…

In a @WSJ story on the U.S. government supporting @intel, CSET Senior Data Research Analyst @jacob_feldgoise states: “From a technology perspective, it really seems like Intel is gaining credibility and confidence.” wsj.com/tech/the-white…

CSET Research Fellow @SamBresnick tells @WSJ’s @NorthropKatrina that it will be a big deal if #China can regularly recover & reuse parts of its rockets: “It pretty radically decreases the cost of building out these giant #satellite constellations that China has made very clear that it wants to build out.” #SpaceRace wsj.com/tech/china-suc…

Watch the rest of the panel, which took place during the 2026 @aspenideas Festival: cset.georgetown.edu/article/helen-…

Using @AnthropicAI’s Mythos as an example, CSET Executive Director @hlntnr describes why a position at the leading edge of #AI development is imperative for U.S. national security. #AspenIdeas

When describing how #AI can impact speed in warfare, @JosephOChapa highlights a point from @EmmyProbasco’s report on the Maven Smart System (MSS): Army leaders hope to leverage MSS to make 1,000 targeting decisions in one hour. Learn more: cset.georgetown.edu/publication/bu…

A website called FLARE-AI has been set up by a group of researchers to report and track #AI harms. Jessica Ji, a senior research analyst at CSET, told @WIRED’s @willknight “it’s a really good initiative” and that she’s in support of anything that makes AI more transparent. wired.com/story/flare-we…

DEADLINE APPROACHING: We’re hiring several research analysts! If any of these issue areas interest you, be sure to submit your application by the July 6 deadline: ⭐ Frontier AI ⭐ Cyber/AI ⭐ Military Applications of Emerging Tech Apply today ⬇️ cset.georgetown.edu/job/research-a…

CSET Research Fellow @SamBresnick spoke with @politico’s @aarontmak about the implications of Chinese companies releasing AI models that are cheaper than proprietary models developed by companies like OpenAI, Anthropic, and Google. #AIRace politico.com/newsletters/di…

Chinese companies are launching AI models that could rival the top U.S. models. CSET Research Fellow @SamBresnick told @CNBC’s @samantha_subin that these recent developments are “a pretty good wake-up call.” #AIRace cnbc.com/2026/06/30/whi…

6 days left to apply to be a CSET Research Analyst! This is an amazing role where you get mentorship & guidance but also credit & independence. Looking especially for folks to work on frontier AI, cybersecurity, and China + military AI. Closes 7/6. cset.georgetown.edu/job/research-a…

Attending @aspenideas this week? If so, be sure to check out @hlntnr's session on U.S. AI competition with China at 1pm local time tomorrow! Details and a post-event recording available at the link below: aspenideas.org/sessions/the-a…

The role of the Chinese Academy of Sciences in the country's S&T ecosystem does not end in the lab. @HannaDohmen spoke with @thewirechina about the extensive support CAS provides tech companies beyond just research.

"We define AI development jobs as roles that require specialized knowledge, skills, and abilities and directly contribute to the technical development of AI systems." Our new report has more ⬇️ cset.georgetown.edu/publication/id…