New newsletter! CSET's Alex Friedland wrapped up all the latest AI & policy news: 1️⃣ The 🇺🇸 AI Safety Institute's rebrand 2️⃣ Meta’s Scale play 3️⃣ Blackmailing, narcing, self-exfiltrating Claude 4️⃣ OpenAI’s $200M DoD contract And more - check it out 👇 cset.georgetown.edu/newsletter/jun…

Oh damn this doll's not fucking around

CSET's Alex Friedland wrapped up the latest AI & policy news in July's newsletter: 1️⃣ Nvidia H20s are back on the menu for 🇨🇳 2️⃣ AI training legal wins for Meta & Anthropic 3️⃣ 🇪🇺 releases AI compliance guide 4️⃣ DOD contracts for AI labs And much more! cset.georgetown.edu/newsletter/jul…

Meta will not sign the EU code of practice for general-purpose AI models.

Guess they'll be revisiting those Grok contracts then

My apple watch is freaking out because I haven't stood up all day. Check out why:

I organized every recommended policy action in the AI Action Plan by the Agency/Department responsible for implementing it... 🧵 (1/5)

Oh guess not

Now updated after Trump signed 3 executive orders operationalizing some of the plan's key recs: 1⃣ Accelerating Federal Permitting of Data Center Infrastructure (bit.ly/3TYCxkh) 2⃣ Promoting the Export of the American AI Technology Stack (bit.ly/3IHm4i5) 3⃣

The new AI Action Plan touts the strategic importance of open models. That's not news to China, where developers like DeepSeek & Moonshot AI have made the PRC the leader in open-weights AI. In Foreign Affairs, CSET's Owen J. Daniels & Hanna Dohmen explain how the U.S. can respond.

If the United States’ AI strategy fails to adequately account for open models, Washington will risk losing “leverage in emerging technology diplomacy in key regions of the world,” warn Owen J. Daniels and Hanna Dohmen. fam.ag/3TZ4CI7

In Nikkei Asia, I argue that, in contrast to their support for Ukraine in 2022, tech companies may hesitate to support Taipei in a crisis due to their economic exposure to Beijing. Unlike Russia, China is central to many of these companies’ supply chains and strategic ambitions.

I wish Anthropic would just set a monthly limit like OpenAI did with o3

CSET's Alex Friedland recapped a massive month for AI news in this month's policy.ai: 1⃣ GPT-5 2⃣ Trump's AI Action Plan 3⃣ The 🇺🇸 gets a 15% cut of chip sales to 🇨🇳 4⃣ Parsing "AI psychosis" And more. Check it out and subscribe: cset.georgetown.edu/newsletter/aug…

I wanted to provide more color on this because its a bigger deal than the main Ai2 account is hyping it as. For example, the entire annual budget of the NSF for AI in 2026 is $655M dollars. To commit to a training models on single line item of about 20% of that is a huge deal.