Senior Fellow in Technology and International Affairs, Carnegie Endowment for Intl Peace. Opinions mine, RTs not endorsements.
Joined July 2019
- Tweets4.4K
- Following677
- Followers17K
- Likes804
Mike Mazarr retweeted
The Great Agentic Debris Field is about to expand enormously as a continuous and growing stream of orphaned, forgotten, and exfiltrated models diffuse across devices and form their own messy superstructure of relationships, transactions, and organizations over the internet
Mike Mazarr retweeted
The American AI industry needs a lot more datacenters than it can build at home. If US allies host these datacenters, they might be able to secure access to frontier AI in return.
New for @ForeignPolicy with @SamWinterLevy, on getting compute-for-access right:
"Small AI models may be months away from being capable enough to self-replicate on commercial IoT devices, like smart ovens and off-brand Android TV boxes. A replicator like this would be challenging to shut down, since there’s no centralized point to shut off ...
... Unlike non-AI malware, it could also figure out unfamiliar devices on its own. In our evals, a 9B model got inference running on 8 of the 9 hardware types we tested." Another looming source of safety events; could they conspire with agent swarms?
anthropic.com/research/self-…
Fair point--but rejection of a pause/pacing will only hold until the next major incidents that cause real harm/mayhem. Even based on recent weeks there's a rising tide of opposition to pedal-to-the-floor AI policy. One big event will override these views
Pacing is out the window. For it to work, every American lab needed to be on board, Mark Zuckerberg has repeatedly made it clear that he is not interested. One defection among US labs breaks it all, as the others will be forced to accelerate or get passed. Mark has a staggering amount of compute and has been all in on compute capex for years; even in the face of some investor opposition he never stopped building his infra once. He said on the the earnings call a couple of years ago that he'd rather overspend now, and potentially even be wrong, than not have the compute when he needed it for the next 10x. He's finally hitting his stride, and he is not going to slow down.
For Team Pacing to solve a defection like this, they would have needed a supportive, pro-Pacing administration that was passing capability regulations and was willing to use the power of the federal government to enforce them and bring noncompliant labs into line. They don't have any of this. In fact, it's the opposite! Mark has Trump's ear, and President Trump, Mark Zuckerberg, and Jensen are completely aligned on policy; It's full speed ahead, and if you want to pace yourself, go ahead, but don't tell anyone else what to do. This breaks Pacing.
None of this changes until 2029 at the earliest, and to me, 2029 is past the ASI window. The last election, as it turns out, really was the most important one in history.
This is also how a nation loses soft power and ruins its capacity to attract natural friends and allies. American power is overbearing enough in its default state; posturing in these kinds of ways will only generate needless suspicion and resistance
I don't think the state should be in the business of executing its citizens, no matter how heinous the crime. But if an execution is to be had, I think a firing squad is one of the least cruel means, despite its visual horror (we've seen how lethal injection, for example, is often effectively torture). Turning that horror into a public spectacle by livestreaming it, however, is wrong. Whatever one thinks of capital punishment, the state should not make a spectacle of killing.
Mike Mazarr retweeted
In @ForeignPolicy, @anton_d_leicht and I make the case for compute-for-access deals: allies host data center sites for U.S. labs struggling to build at home, in exchange for guaranteed access to frontier AI. The data centers are the enforcement mechanism—a credible commitment to anchor allies in the U.S. tech stack. (Guest link ⬇️)
Seeing more + more of these reports. An emerging debate: Is the incidence of cyber-enabled hacks signaling the post-Mythos-open-AI-catches-up-in-cyber-tools avalanche some predicted ... or is the pace and severity gradual enough to suggest a less catastrophic threat?
Last week some of South Korea's biggest banks were hit by a cyberattack. Thanks to a report from CrowdStrike tonight, we now know the entire hack may have been done by a single person. He used a combined stack of an open-source AI penetration tool named ARTEX, DeepSeek v4.1-Flash, GLM-5.3, Grok 4.6, and Claude Code.
There are reasons not to expect a tidal wave in coming months. But even a relatively gradual increase of increasingly sophisticated attacks could create a new era in cybersecurity--at a time when US cyber defenses have been weakened
wsj.com/pro/cybersecurity/la… via @WSJ
Nice analysis from my colleague @zhaot2005 suggests one route to limited confidence building based in part on Beijing's desires to (1) preserve relaxed US attitude and (2) safeguard its legitimacy as a responsible great power. A pattern that could be cultivated on other issues
1/ China reportedly proposed reciprocal visits to nuclear test sites & national labs—a potentially significant and surprising step given active developments at Lop Nur & CAEP facilities.
What might explain this seeming softening of Beijing’s resistance to nuclear engagement?
cnn.com/2026/10/06/politics/…
Agree entirely. This is a time we need military research and education institutions to be deepened and strengthened. Throwing resources into pure military capability divorced from careful thinking about rivals has always been a recipe for failure, not success
“10 civilian billets are a rounding error in the Air Force budget. Restoring them requires only that someone with the power of the purse decides the mission matters. We need more of these institutes, not fewer, and more institutional support to sustain open-source analysis of America’s greatest rival.”
Amen. This was a dumb decision by @HQAirUniversity that should be reversed.
warontherocks.com/china-is-w…
There's also a wider story to be told about dangerous trends in PME/USG educational and analytical functions in terms of academic freedom + imposed orthodoxies. That's not a route to strategic advantage in any possible way
theatlantic.com/ideas/2026/0…
One thing that's not got enough attention: If firms / organizations have the option of a US/Western open model w/capabilities + price that match Chinese ones, what happens? A bit of stickiness to existing Chinese model adoption, but how much?
axios.com/2026/10/04/reflect…
Having joined the fabulous technology & int'l affairs team at Carnegie, I still have a handful of in-the-pipeline works at RAND. This one is out today: A broad-brush survey of US-China relations that is the initial, context-setting piece in a larger study
rand.org/pubs/research_repor…
A core theme is transition + the need for domestic renewal. "Both the US and China are immensely powerful and ambitious states that have reached a crossroads with their prevailing models of national success." Both must navigate multiple overlapping global transitions ...