My report, “Small Force, Big Data: Defense AI and New Zealand,” is out: https://t.co/ScnccckE5x. I explain that defense #ArtificialIntelligence isn’t optional - it’s the difference between sovereign capability & strategic irrelevance. Thanks @HeikoBorchert. #nzpol#NewZealand
New working paper: Maritime Vulnerabilities and Geopolitical Disruptions in Global Shipping
In a new paper with @SimonFuchs4 and @WoanFoongWong, we quantify maritime vulnerabilities around the world.
We combine novel data on shipping routes, vessel characteristics, and international trade to measure countries' exposure to maritime chokepoints, their ability to reroute, and their dependence on foreign vessels.
Motivated by this evidence, we develop a quantitative spatial model of global trade with endogenous shipping markets. The model allows us to quantify how disruptions affect shipping routes, freight rates, and economic welfare across countries.
We find that the busiest shipping routes are not necessarily the most vulnerable. What matters is how easily ships can reroute and how shipping markets adjust. Vulnerabilities differ substantially across countries, with U.S. exposure spread across distant chokepoints and China's more concentrated in East Asia.
Finally, we study two geopolitical scenarios: disruptions to maritime routes in East Asia and restrictions on Chinese-built or Chinese-operated ships. We quantify their economic consequences, including spillovers to third countries.
Paper: https://t.co/8gGdUT9oAU
BREAKING: Google warns that state-backed hacking groups linked to Iran, China and North Korea are increasingly using AI to conduct cyber espionage, steal sensitive information, infiltrate organisations and develop sophisticated attacks. Separately, Russian-speaking hackers have been identified in what Google describes as potentially the first fully automated AI-powered cyberattack.
In its latest threat intelligence report, Google details how these actors are moving beyond simply using AI to write code or assist with research. They are integrating large language models (LLMs) and autonomous AI agents into multiple stages of their operations.
RUSSIA: 🇷🇺Google identified Russian-speaking hackers who used a network of AI agents to independently discover software vulnerabilities and develop exploits without human intervention. Their operation targeted French websites in the media, entertainment and education sectors. Google says this may be the first observed case of a fully automated vulnerability discovery and exploitation process.
CHINA: 🇨🇳Chinese-linked cyber espionage groups used AI during the third quarter of 2026 for reconnaissance, scanning, researching targets and weaponising software vulnerabilities. Google also identified a Chinese contractor known as LEAKY CASTLE using LLMs to analyse, translate and extract intelligence from stolen data. Other Chinese-linked actors abused AI platforms and file-hosting infrastructure to stage malicious files and exfiltrate stolen documents.
IRAN: 🇮🇷An Iranian government-backed group known as CALANQUE ION, previously tracked as APT42, has been using LLMs, including Google's Gemini, to support reconnaissance and targeted social engineering.
NORTH KOREA: North Korean actors are using AI to generate fake CVs, prepare for job interviews and impersonate Western IT professionals to secure remote employment. Google also identified operations targeting cryptocurrency companies in North America and Europe, involving fake recruiter profiles, AI-generated identities and fraudulent meeting platforms used to compromise victims and gain access to internal systems.
Google further warns that attackers are increasingly exploiting AI infrastructure itself, including autonomous execution capabilities and coding tools, to move deeper into compromised organisations.
The findings point to a significant evolution in cyber warfare and espionage.
Ahead of the Trump–Xi–Putin trilateral at APEC Shenzhen: this will be the clearest test yet of my concept of "Triangle of Happiness", the functional triumvirate of realpolitik between the three nuclear powers with the deepest fingerprints on the two proxy wars, Ukraine and Iran.
First takeaway: We are not in WW3 and this is not Yalta 2.0: no carve-up into strategic spheres of interest (Western Hemisphere for the US, Europe for Russia, the South and East China Seas for China). We are at the very beginning of this new strategic competition, not its end. And we know that the US wants to emerge victorious, so do China and Russia.
Second takeaway: This is not 19th concert of global powers either. Middle powers bridge the two systems (security with the US, economy with China), hedge against all (India) or balance as hubs through geography, tech leverage or raw materials as they do not accept the bifurcation of the global system and do not want to get caught in a new binary world.
Third takeaway: It is a new Cold War of global grayzone warfare. The US wants to win the AI-steered 4IR and establish a Pax Americana 2.0. China and Russia want to prevent it and build an order on their own terms, structures and formats. Most of the competition takes place below the treshhold of a conventional and nuclear exchange between the US and the DragonBear as the modus operandi of strategic coordination between China and Russia.
Fourth takeaway: Iran is the most acute file as it has triggered a fourth systemic risk-driven global crisis through the destruction of the critical infrastructure in the Gulf and the closure of Hormuz. China and Russia sustain Tehran, so any durable deal runs through them. Two proxy wars plus the greatest chokepoint disruption - this might result in this fourth systemic risk after 2007, 2020 and 2022, what I define as a Global System Rupture, a possible fourth systemic global crisis. The Iran war can flare again within weeks.
Fifth takeaway: Russia’s logic is unchanged. In December 2021 I outlined the three dimensions of Russia’s comprehensive war: subjugating Ukraine, destroying the European security order, and taking pole position as the wild card in the US–China rivalry. Putin went to war precisely to put Russia on the same level as Washington and Beijing. A three-way photo in Shenzhen is part of that payoff, even though Ukraine is officially off the trilateral agenda.
Sixth takeaway: Each side comes to Shenzhen for something different:
- Russia drives the format. It wants global status and the image of strategic parity with Washington and Beijing.
- China hosts and gains prestige. It sets the terms, avoids owning any failure, and still rejects three-way nuclear limits.
- The US wants an opening on arms control now that New START has expired, and it wants help on Iran. Trump arrives weakened by the midterms and in a hurry to make deals before the new Congress sits in January
- Unexpected wild card could be Korea talks and setting the stage for a political settlement of the armistice between North and South Korea within the Triangle of Happiness.
Seventh takeaway: The triumvirate’s task is minimal: agree on transparent rules of engagement that prevent kinetic or nuclear war (CSBMs, direct channels of communication, etc.), keep the global system afloat, and manage the cascading risks. Other files can’t be ignored either: the AI revolution (openly a front of the new Cold War since the Nvidia affair), nuclear proliferation, and keeping humans in control of nuclear weapons.
Bottomline: Managed confrontation for the sake of the strategic stability amid MAD 2.0, not any grand bargain, separation of spheres of influcence or Yalta 2.0. Stay tuned.
.@paulfchristiano on what the end output of the intelligence explosion (running within a datacenter) will be:
Billions of minds, thinking 10s of times faster than humans, learning from all their parallel experience, with better learning algorithms than the ones evolution gave us.
Excited to see @jamescrabtree's new book, The Perimeter, come out Jan 12. The book blends history, on-the-ground reporting and interviews with strategists & practitioners to explain what’s at stake along the first island chain. It’s sure to be a must-read! https://t.co/YFAHeno64R
Dwarkesh has MAJORLY shortened his timelines: ASI is "possibly only a couple months away"
Just a few months ago, he thought it was the "better part of a decade" away.
"If I had to pick my median year for when automated researchers will be as good as the very best human AI researchers (and assuming the labs don’t intentionally slow down), I’d guess 2028."
"I’ve now come to the view that we’re only a couple of years away, and possibly only a couple of months away, from AIs that are as good as the best humans at AI research. I’ve also come to believe that automating AI research will compress so much progress into a single year that, soon thereafter (so by 2030, but plausibly as early as next year), we’ll have AIs that are at least as good as the best human experts in every field. I know this is a wild claim, and it’s one that I myself didn’t believe even a couple of months ago.
It’s crazy how fast we went from, in 2023, “This model can’t reliably add four-digit numbers,” to, in 2024, “This model can barely solve any problems in a high school math competition,” to, in 2025, “Oh, it can win a gold medal at the International Mathematical Olympiad,” to, earlier this year, “Wow, it’s solving open Erdős problems, but maybe just because no human has attempted to solve these particular problems, and a similar solution was already in the literature,” to, about a month ago, “Okay, it fucking solved a Millennium Prize problem in 88 hours.”
Some of the most well-informed researchers I know, people who have had a very good track record of prediction over the last few years, expect AI to beat the best humans at AI research as early as next year. And 2027 is also the straightforward implication of plugging METR’s finding—that Anthropic’s rate of AI progress has doubled in a year—into a takeoff model.
As former guest Ajeya Cotra put it:
"If we take it at face value that AI is now nearly doubling the rate of AI capabilities progress and assume acceleration from AI is smooth, this naively implies that we would reach extreme superintelligence within a year."
Today’s AIs already do work at about 10 times human speed. And automated researchers would also be working around the clock, nights and weekends included. As a result, each individual automated researcher will be cramming in roughly 50x more work every calendar week than a human researcher can do.
And this is before we account for the fact that, in this scenario, every automated researcher will be at least as good at AI R&D as the very best humans.
My sense is that around 80% of the algorithmic progress at a given frontier lab is driven by its top 50 researchers.
In the last four years, we’ve gone from a preschooler to a savant intern. If you made another jump that big, but starting in 2027 or 2028, you’d probably get a mind that would be vastly superhuman in domains that are even slightly verifiable.
For example, it’d be better than Nvidia at coming up with the next generation of AI accelerators. It would make more money than the current human versions of Citadel and Jane Street (if it had equal access to the market). If you gave it remote access to a materials science experiment lab, it’d be better than the rest of humanity at finding a room-temperature superconductor.
*** AISM note: the whole blogpost is really good and worth reading.
New Zealand’s politics appear unstable after national elections failed to produce a clear winner last week, writes expert @JoshKurlantzick. “At a time when the country faces severe economic challenges, policy paralysis is a recipe for disaster.”
“No matter who takes power, they will have a weak government. This is the last thing New Zealand needs now,” he argues.
The country “already faces a brutal combination of economic challenges, all made worse by the Iran war and its effects on global energy markets. The cost of living in New Zealand is high, due to its isolation from the rest of the world, import dependence, and expensive house construction.”
“It could be time for New Zealand to rethink its political system and make major adjustments to it, lest its supposed idyll continue to fade economically, to the point that it resembles far poorer democracies, like Greece, than former peers like Australia,” Kurlantzick writes.
My latest essay with @akhil_oldsoul, in which we explain that we’ve moved from the Unipolar Moment to the Unipolar Sunset. And no, this isn’t an argument about “American decline.” Read it here to find out more:
Cc: @PacificForum
https://t.co/4UJD0u0xhS
Quick little hot take on Jensen Huang and the honey pot of the century
Also...why can't rich powerful men just keep it in their pants? Is it really so difficult?
https://t.co/wUFaWG6eVr
Fable 5.2 = timelines for superintelligence are collapsing - some experts think (and are panicking) that it comes in 2027. The US will be first. China, meanwhile, steps up exercises practising blockading Taiwan to delay/cut off the chips the US needs for superintelligence?
"This trajectory, if continued, means a world-changing intelligence explosion in 2027, so the public needs to know."
Every chart right now is a screaming red alert.
🟧 For the first time ever, Sentinel has issued an orange alert, due to the METR report.
Sentinel (@xriskfyi) is a weekly report, written by the world's top forecasters, that monitors global risks.