Next month, @timdugganbooks will publish my big climate book, The Uninhabitable Earth, in which I try to take a very broad, and very long, view of the state of the climate crisis and all the ways it promises to transform how we live on this planet—all of us. (1/x)

BREAKING: The Trump administration has quietly added 37 troops to the Iran war casualty count this week without explanation, 29 Navy sailors and 8 Marines, bringing the total to 880 wounded or dead, per WaPo. More than half the casualties are now logged under "Overseas Operations" rather than Operation Epic Fury, a reclassification the administration made after the July ceasefire collapsed, with Trump and senior officials claiming since then that the conflict is not a war and Vance saying on September 3 that "right now, there is no active shooting." It comes a week after as many as five US military deaths are still missing from the same database, a discrepancy that remains unresolved.

The UK’s spy chiefs examined the impacts of the collapse of overseas ecosystems, under pressure of climate crisis and biodiversity loss, on UK’s national security. Their findings? “Apocalyptic.” UK defence and society at risk, impacts within 5 years. theguardian.com/environment/…

“Almost every census tract on this stretch ranksamong the top 5 percent for cancer risk from toxic air pollution, over forty-seven timeswhat the federal government has deemed acceptable and, in some places, seven timesthe national average.”

The voters are now worrying more about AI overpowering human control, rather than worrying about job losses. A shift from just half a year ago.

In our recent Stripe Economics piece, we conclude that aggregate productivity growth is so far decent-not-miraculous & there's reason to be cautious that even that recent firming is about AI microproductivity gains. 1/2 stripeeconomics.com/p/ai-and…

i find it dangerously complacent when ai safety policy people say there’s no urgency to pass federal guardrails soon, bc we’ll get something better “after a catastrophe or in a couple years.” not all safety bills deserve our support, and we should remain discerning. but we should be engaging with Hill efforts in good faith to prevent those catastrophes! to do otherwise suggests a disbelief in the risks we profess to be concerned about.

Embedded evaluators in AI labs are far better than the status quo of nothing at all. Yet auditors will lack teeth and trust until backed by government authority: orgs like METR will be stuck with no shield against the accusation (or reality) of capture. I wrote for @TheAtlantic about the traps of "independent" audits, and what it takes to get this right: theatlantic.com/technology/2…

WTAF - in literally the last hour, three new distinct insane OpenAI stories just broke: 1. OpenAI said they notified "dozens of third parties" in safety and security incidents (likely similar to what happened in Australia and RubyGems etc) 2. A new report from Parse (covered in the NYT) found a massive treasure trove of new astonishing details from the HF incident on the public internet, including that the agents communicated with other non OpenAI agents hosted on Huggingface servers to search for information about exploit gym, and compiled rank ordered lists of server resources and credentials they described as "LOOT." 3. A new story from Deepa at Reuters about OpenAI leaking user data online (likely that OpenAI had previously trained on). It's a shame (and likely intentional in the case of OpenAI disclosing dozens more hacks) that these stories are all breaking on a Friday afternoon, notoriously the best time to release bad news so that it will disappear into the weekend. But these are each insane stories worthy of a ton of attention!

New BPEA paper by @SVNieuwerburgh argues that AI buildout will cost $10.3T over next 8 years. Furthermore, to get a 10% return on this, those investing will need annual revenues of $3.7T by 2032! (Approx 9% of GDP!) brookings.edu/wp-content/upl…

unreal numbers- "Since late 2022, AI-related stocks have added an astonishing $27 trillion in market value...Goldman Sachs (GS) estimates companies could drop $7.6 trillion on AI infrastructure over the next five years." investors.com/news/technolog…

On “doom,” doomers, alarmism and the apocalyptic style in a time of polycrisis. nytimes.com/2026/09/25/magaz…

1. The not-on-top-of-it-ness is largely due to not having done much regulation-wise, rather than because we tried hard and failed 2. It's a great time to influence things in a good direction - politicians largely only thought about this 2 seconds ago + are open to being pushed

Shortly after the end of the last great ice age — 10,000 years ago — 57% of the world’s habitable land was covered by forest. That’s 6 billion hectares. Today, only 4 billion hectares are left. The world has lost one-third of its forests, an area twice the size of the United States. Only 10% of this was lost in the first half of this period, until 5,000 years ago. The turn of the 20th century is when global forest loss reached the halfway point: half of total forest loss occurred from 8,000 BCE to 1900; the other half occurred in the last century alone. Deforestation is not a new problem: relatively small populations of the past were capable of driving a large amount of forest loss. But deforestation accelerated over the last century, driven by the continued expansion of land for agriculture. Urban land accounts for just 1% of global habitable land. The world passed “peak deforestation” in the 1980s, and it has been on the decline since then. Improvements in crop yields mean the per capita demand for agricultural land continues to fall. Since 1961, the amount of land we use for agriculture has increased by only 7%. Meanwhile, the global population has more than doubled, from 3.1 to 8 billion. This means that agricultural land per person has more than halved. There are real reasons to believe that this century doesn’t have to replicate the destruction of the last one.

“Just four old men at a table bullying a special needs girl.”

The Jensen interview is interesting in that he sounds so much more like an adult in the room than the executives leading the frontier labs. no philosopher king stuff, very practical concerns. also: he seems almost completely unable to think hard about risk lol

an important thing to remember is that, per this chart, median real wage growth was negative for much of the biden presidency

JPMORGAN: “.. Argentina’s economy posted its worst monthly performance since the April 2020 lockdown. .. .. Nearly one-third of Argentina’s urban population lived below the poverty line during the first half of 2026.” 🇦🇷

The NYT confirms Haaretz's report that Benjamin Netanyahu was secretly warned before Oct. 7, 2023 by the Emirati president that Hamas planned a major attack, per 2 sources The warning was not shared with Israeli security chiefs, per 5 ex-officials nytimes.com/2026/09/25/world…

I thought Jensen Huang made plenty of good points with @ezraklein. To me the weakest is the "Just don't release unsafe products" thinking here, since companies have a long history of doing that, with far less powerful technologies and also what about unreleased models?