The Center for a New American Security develops strong, pragmatic, and principled national security and defense policies. Join us: cnas.org/join.
Washington, D.C.
- Tweets 38,862
- Following 1,935
- Followers 64,503
- Likes 6,630
"Most people I talk to at these AI labs are just genuinely scared about the direction these AI models are heading," shares @janet_e_egan.
Read her full comments on the breach of an Australian government service by OpenAI agents in @politico.
politico.com/news/2026/09/27…
156
The surge in global demand for Chinese AI could generate significant problems for the U.S.
“The ultimate concern is that the integration of Chinese AI models pulls countries into a Chinese technology sphere of influence that hardens into geopolitical alignment,” @danielremler tells @CNBC.
cnbc.com/2026/09/26/china-ai…
1
1
3
1,067
The U.S. has long had the tech edge on the battlefield. @paul_scharre says that edge may be eroding faster than we realize.
He joins @onedecisionpod to break down cheap drones, military AI, and why the offense-defense cost balance is shifting fast.
1
2
965
The Trump-Xi meeting produced limited outcomes on trade issues: a two-month extension of the Busan truce, including the pause on China's rare earth export controls and lower tariffs on a $30B batch of non-sensitive goods – with both sides pushing a broader trade deal for next year.
Here is what our research says about why this keeps happening 🧵👇
1
1,097
📈 In "Hit It with Your Best Shot," @EmilyKilcrease1 argues that truce extensions are tactical moves, not strategies.
The U.S. needs a clear doctrine for economic pressure to achieve favorable long term trade outcomes: cnas.org/publications/report…
1
1
1
235
🌏 And if at the end of the two months there's still no truce?
"Whither the Coalition" examines whether the U.S. could actually build a sanctions coalition against China, and finds even close allies are far from committed: cnas.org/publications/report…
178
Mechanisms like export controls used to be "seen as national security issues, and that’s not something for trade negotiators to haggle over," says @EmilyKilcrease1.
Read more on how this approach has shifted under the current administration in @globeandmail.
theglobeandmail.com/business…
611
CNAS retweeted
The first post on my new Substack. Check it out: rhfontaine.substack.com/p/a-…
9
16
6,596
We're not just on X, come find us on Instagram too: instagram.com/cnasdc/
2
1,994
CNAS retweeted
The Trump-Xi summit should be viewed not as a one-off but rather as another diplomatic spike in a long year:
1. The state visit was the second of four U.S.-China summit opportunities in 2026. And even more than last time, this one was more sizzle than steak.
2. Heavy on the pageantry, there were more military flyovers than substantive agreements. The two sides extended the trade truce, but only by two months, reduced tariffs on $30B in goods, and announced an AI dialogue.
3. For China, the visit was a major opportunity to show the world - and the Chinese people - that Xi and Trump are two superpower leaders, and that the Chinese leader is now recognized as such. Hence the U.S. literally rolling out the red carpet, top CEOs joining him in Washington, the President greeting his plane, etc.
4. There was naturally wall-to-wall coverage in the Chinese press. The summit served Beijing’s narrative of an inexorably rising China.
5. The U.S. got two pandas, which is something, I suppose. It seems to have gotten nothing on Taiwan, the South China Sea, gray zone activity, Iran, Russia-Ukraine, AI distillation, human rights, or other grievances.
6. Instead, stability was the watchword. Both sides emphasized commonalities and abstract goals rather than the reality of strategic competition or the very real divergence in outlook.
7. Xi even invoked history to remind Americans that they once teamed with China to fight Japanese militarism. Perhaps, he suggested, we’re seeing echoes of that era today.
8. That perturbed some Japanese experts. Perhaps, if Xi repeats often enough that the U.S. and China were once united against Japan, Americans could be convinced to do it again? No chance. The Chinese and Japanese governments were very different back then, to put it mildly.
9. No one today confuses the two or believes that Japan, rather China, represents America’s great long run strategic challenge. Even Trump, despite his affection for Xi and his 1980s mindset on Japan, says the U.S. needs to beat China on AI.
10. It’s good to keep up bilateral communications, given the power of both sides and the stakes in the U.S.-China relationship. Rather than simply look ahead to the next opportunities in Shenzhen and Miami, however, Washington should focus on everything that happens in between the meetings.
11. That means returning to vigorous competition, strengthening the U.S. position in all ways, working closely with American allies and partners, and constraining key Chinese capabilities. We have a long way to go in reassuming a strategic mindset.
2
10
25
4,869
Add your email to the list here: cnas.org/weekend-reads
233
"The day has been filled with so much pageantry that there hasn't even been all that much time for the two presidents to sit down and work out their differences—and I actually think that's by design."
Hear more from @Liza_D_Tobin on the Trump-Xi summit on @business.
bloomberg.com/news/videos/20…
1
2
1,060
"If you shoot down a $1,000 drone with a million-dollar missile, you're losing everytime. It's not a winning proposition."
Watch @paul_scharre discuss how drones and AI are transforming war on @onedecisionpod.
piped.video/watch?v=iYlsCirZ…
2
1,260
The lack of prominent Chinese CEOs at the Trump-Xi state dinner is “a rather depressing marker of how bad the relationship has gotten.”
Read @EmilyKilcrease1's comments on Xi's visit to Washington in @nytimes.
nytimes.com/2026/09/25/busin…
922
