🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

In just one hour: watch online or join in person the Taking Forward Nuclear Risk Reduction Side Event. Institute for Security and Technology is thrilled to partake in this event, with United Nations Institute for Disarmament Research, BASIC, and the Govts of The Netherlands and Switzerland. unidir.org/events/npt-sid…

The event will discuss how to further nuclear risk reduction, taking advantage of new political and technological opportunities. Institute for Security and Technology’s L. Walker will take part in the discussion, touching on nuclear crisis communications and the role they can play in reducing nuclear risk.

On crisis communications links: "As critical as communications are to ending a conflict, it is often the first thing to be targeted at the outset of one. If there are not robust avenues in which to negotiate, you may find yourself going without." L. Walker #NPTRevCon

Last Monday Institute for Security and Technology’s L. Walker joined a #NPTREVCON side event hosted by United Nations Institute for Disarmament Research, BASIC, and the Dutch and Swiss governments focused on taking forward risk reduction. media.un.org/en/asset/k1y/k…

Over the last month, team IST participated in #NPTRevCon, joining international efforts to reduce the risk of nuclear conflict. 1/8 securityandtechnology.org/blog/ist-at-th…

The NSS mentions enhanced crisis communications with China - good! But it's also been a longtime priority that has thus far failed to have regular impact. Check out which 🇺🇸🇨🇳 links have been attempted in L. Walker and my "Atlas of Crisis Comms": securityandtechnology.org/wp-content/upl…

A new piece from Alexa Wehsener and I, where we discuss the implications of the new Office of Strategic Capital:

The Office of Strategic Capital Department of Defense 🇺🇸 was established this month to boost defense-tech investment. IST’s L. Walker & Alexa Wehsener write, “the true test of the well-intentioned office will be whether it has sufficient impact on the American innovation ecosystem.”

📣 NEW: With support from Department of State's Bureau of Arms Control, Verification, and Compliance, we're releasing the product of a year's worth of research into the strategic stability risks posed by integrating AI into NC3. securityandtechnology.org/virtual-librar…

New from BRSL's Andrew Reddie : "Why the alarm over Russia’s use of hypersonic missiles in Ukraine is misplaced". Dr. Reddie examines the two types of alarmism in the response to the recent Russian use of hypersonic Kinzhal missiles in Ukraine. Read it now: thebulletin.org/2023/03/why-th…

Can't wait for this conversation with some of the best and brightest in the #AI-#cyber and #governance communities from CISAC & on Bluesky @stanfordcisac.bsky.social Google OpenAI Paul Jankura on the sidelines of RSAConference and sponsored by Berkeley Risk and Security Lab and United Nations Institute for Disarmament Research. Join us on April 24!

As the G-7 convenes in Hiroshima this weekend, I wrote a piece for Foreign Policy about the city's enduring message—and the country's challenging security environment. foreignpolicy.com/2023/05/19/g7-…

Thread 📜 OTD in 1963, the US and Soviet Union established the "Washington–Moscow Direct Communications Link" or hotline. At the time, the @Guardian reported that this step would be a "small, but useful, measure of arms control" especially in light of the #CubanMissileCrisis.

Bigger isn’t better, but major AI policies (like the Biden EO) focus only on model size & compute. However, models are useless without data! In our latest work, we highlight the need for data-centric AI policy and reveal the shortcomings of the status quo. arxiv.org/abs/2409.17216

For Carnegie Endowment I wrote about the perils of nuclear ambiguity, and what Iran (and the world) can learn from the saga of Iraq's nuclear program. TLDR: playing political football with international oversight rarely ends well for anyone: carnegieendowment.org/research/2024/…

A few takeaways from my recent piece for Carnegie Nuclear Policy about the risks of a more opaque Iranian nuclear program (and some cautionary lessons from the Iraqi experience)

China is bypassing U.S. export controls on GPUs, but how? In my latest paper, "Whack-a-Chip," we provide the first public evidence of how China is using advances in ML to make old chips useful for SOTA results. 👉🏽 arxiv.org/abs/2411.14425 Andrew Reddie Berkeley AI Research Berkeley Risk and Security Lab

Delighted to see the latest from Berkeley Risk and Security Lab w/ Ritwik Gupta 🇺🇦 and L. Walker in the wild. Using the recent release of Tencent's Hunyuan-Large model, the paper begs the question as to whether hardware-oriented #AI #export controls are doing what policymakers want...

For Foreign Policy I wrote about Iran and why nuclear uncertainty is bad for regional stability (link below). Unclear diplomatic priorities + backlash against the IAEA w/in Iran are taking us down a dangerous path.