Today’s briefing:
— Should we slow AI?
— A pre-modern smartphone
— Why this embassy just closed
Your Insider’s briefing:
— Should we slow AI?
— A pre-modern smartphone
— Why this embassy just closed
Good morning {{first_name | Intriguer}}. I’m reading the brilliant Infinity Machine biography of DeepMind founder Demis Hassabis, which casually notes that decades ago, futurist Ray Kurzweil was already crunching the computation numbers behind Moore’s Law to extrapolate that “we’ll have effective software models of human intelligence by the mid-2020s.”
So here we are smack bang in the mid-2020s, exploring why everyone suddenly freaked out over AI, yet somehow Kurzweil might see nothing sudden at all.
Shall we explore why?
 | Managing Editor Jeremy Dicker |
🎧 Prefer this briefing via podcast? Become an Intrigue Insider today!
Quote of the day
“Sometimes we may have information that is so sensitive, client information that we really want to protect, we don’t want to put it to any cloud vendor”
That’s the CIO of Latham & Watkins, America’s second-largest law firm ($8B in revenues), now buying its own AI servers and fine-tuning its own AI models. It’s an interesting move that could flip the “AI will kill law firms” narrative, and maybe offer a glimpse at the enterprise AI endgame (private stacks + proprietary data).
Blink of an AI.
Anthropic CEO Dario Amodei.
"Gen Z quits job!" is not ordinarily worth a headline, unless they’re leaving millions of unvested equity on the table to warn that we’re all going to die.
So cue the classic domino meme, because this story escalates fast, starting with…
That first domino
Back last Tuesday, Jacob Coxon, a 27-year-old AI researcher at Anthropic and previously OpenAI, announced he was out. And this was a shock because of...
a) what he was losing (life-changing amounts of unvested, pre-IPO equity)
b) what he said (the world's top AI labs are "gambling with our lives"), and
c) the splash this all made (100 million views plus a global debate).
Then just when we were busy yelling about that first domino, we got...
The second domino
Several of Jacob's colleagues actually agreed, whether fellow researcher Evan Hubinger ("we really do earnestly believe AI could kill all humans!") or Friday's shock risk report on, say, the Houthis using Claude to build missiles, Mali's spooks using it to spook, oh and is China's Kimi AI actually just Claude in a hat?!
Then just as we're busy yelling about that, we get...
The third domino
The Anthropic CEO himself, Dario Amodei, dropped his latest manifesto over the weekend, titled 'We Must Pace the Frontier'. It's worth a read but basically argues the AI risks are now so high, labs should start...
a) voluntarily slowing their own advancements
b) embedding independent inspectors to vet models, and
c) coordinating AI safety across the free world to avoid trading safety for speed.
Why now? Amodei cites the summer’s developments in...
i) control, like autonomous swarms causing unintended harm (per the OpenAI rogue that hacked Hugging Face), and
ii) speed: recursive self-improvement means AI is now accelerating itself.
And sure enough, other AI luminaries then quickly chimed in with an amen, including Musk (Grok), Altman (OpenAI), and Hassabis (Google / DeepMind). Four egos of that magnitude all agreeing feels a bit like the planets all aligning and nudging the ancients to reach for a new calendar.
But you know what? There might be an ego that eclipses them all, in the form of...
The fourth domino
On a weekend visit to Ireland, President Trump dismissed them all as "negative forces" warning of "things that won't happen." And he went a step further, tugging on Amodei's free world thread above: "we're leading China on AI... and, frankly, I want to keep it that way because whoever wins AI, wins."
So to round out the domino meme, that’s how you get a zillennial loud-quitting on a Tuesday, and the president of the United States himself — on an Irish fairway — somehow weighing in by Sunday.
Intrigue’s Take
So... what — and we cannot stress this enough — is going on?
The answer hinges a little on your worldview, but there could be three main theories here, starting with just taking these CEOs at…
Face-value: those closest to AI (including insiders in the Intrigue group-chat — join today!) seem…
| Intrigue Insiders | Membership |
Our experienced, unfiltered takes on what's really going on.
| ✓Daily audio edition | ✓Completely ad-free |
| ✓Insider Telegram community | ✓Monthly AMAs |
| Start your free trial → | $14.99/mo · $149/yr 7 days free |
Intrigue’s Take
So... what — and we cannot stress this enough — is going on?
The answer hinges a little on your worldview, but there could be three main theories, starting with taking these CEOs at…
Face-value: those closest to AI seem genuinely rattled by what they've built, and the prospect of humanity now losing control. They actually want government guardrails, because raw market realities mean one CEO tapping the brakes simply means handing the crown to a rival, who then drives us all off a cliff.
That's partly why OpenAI's Altman says he's now delaying his IPO, though there are inevitably takes much more...
Cynical: trillions of dollars in tech egos suddenly all hold hands to sing kumbaya? C'mon. Staring down the barrel of massive legal liability, antitrust scrutiny, and local community backlash, all while China's high-punch, low-cost labs casually gut the entire premise of America's trillion-dollar data centre capex? In that world, the first US lab to now tap the brakes gets wrecked by the market, so they all invent a shared monster to justify slowing together. And absent the capex moat, they're asking DC to provide a new one: regulation.
Meanwhile, those underwriting the capex, whether Japan's Softbank (already down 12%) or America's Nvidia (markets open shortly), get left holding the bag. Altman just doesn't want a post-IPO fade and quarterly earnings calls while this all plays out.
But even if you buy the face-value, the cynicism, or some small-batch blend of the two (that’s us), there's then the...
Geopolitical: these CEOs know (and Trump just said out-loud) that even if the free world voluntarily taps the brakes, their own commercial fears just get swapped for a geopolitical one; so absent some kind of verifiable deal with China, the result is just handing an authoritarian state the reins on a tech that'll write (or end) the next century.
Relatedly, Zuckerberg (Meta) has been unusually silent this week, maybe because he just made this same point last month, warning even a one-month slip “could add significant risk to American leadership while letting foreign models race ahead.”
Anyway, while Trump's response signals that any slowdown would be unilateral disarmament, cynics could wonder whether Amodei et al are really just pushing the president to bless their oligopoly via a mix of natsec and safety rules, say, blocking US access to China's low-cost AI like DC already blocks its low-cost EVs.
And yet there’s a clear through-line across all three theories worth pondering: one lab won’t yield to another; one country won’t yield to another; and maybe human intelligence won’t yield to a synthetic other. In each case, absent real trust or a sovereign above, nobody wants to be decided for.
And yet it’s out of that last human-synthetic refusal that we see grounds for hope: even amid historically low trust, high stakes, and opposing worldviews, surely the prospect of shared annihilation can be enough to bring adversaries to the table?
History says maybe: Biden and Xi already agreed not to cede their nukes to AI. Go back even further, and the Soviets agreed to similar Cold War guardrails around nukes and Outer Space. Humans can figure this stuff out not because we ever trust one another but because we ultimately share the same rank, existential self interest.
For China, joining a global AI re-think might even make tactical sense if — as some suggest — China’s labs are still trailing the US frontier, even by a few months. So why not accept a coordinated pause that prevents that gap from getting wider? An obvious doubt is around verification but again, we’ve figured that out before (nukes).
Ultimately, maybe history has taught us trust isn’t really essential to build the basic guardrails if both arch rivals already agree they’re standing on the same precipice. So let’s see what comes out of the next Trump-Xi summit in DC (September 24), though an initial article out of China’s Communist Party-run Global Times dismisses Amodei’s letter as a Cold War move to curb China’s AI development.
The Global Times can often run its own lines, but this is a bad initial sign.
Today’s briefing is sponsored by:
Up to $1M in identity theft recovery
If identity theft happens, Coveron doesn't leave you alone. Get dedicated case managers, legal support, and up to $1M in recovery assistance. We handle the mess so you don't have to.
One scam can cost you everything - protect yourself now. The first 100 users get 20% off with code beehiivenewsletter.
30-day money-back guarantee. Terms and conditions apply.
[👋 Prefer ad-free reading? Sign-up to Intrigue Insiders today!]
Meanwhile, elsewhere…
🇮🇷 IRAN — The latest.
Oil’s back up around $110 after another rollercoaster weekend: Friday’s briefing hit your inbox with (now confirmed) reports that drones hit Saudi Arabia’s one east-west pipeline that had long curbed oil dependence on Hormuz. Then a rare Iran-Gulf sit-down planned for Oman today got parked — the Iranians say at Riyadh’s request. Meanwhile, NYT reporting suggests Iran’s own hardliners purposefully sank US talks earlier this year. (Reuters)
Comment: Expect oil prices to shoot even higher if Saudi Arabia can’t get its pipeline back online within days (no word yet). Meanwhile, and mindful not to read too much into President Trump’s passing lines, but he just brushed off WSJ reporting that China’s satellite imagery might’ve helped Iran target a strike that killed three US troops earlier in the war — a possible indicator he wants nothing to rock his Xi summit on the 24th.
🇬🇧 UNITED KINGDOM — US yoinks embassy staffer?
US authorities have detained one of their own embassy staffers in London and flown him home on allegations he owned child abuse material. (BBC)
Comment: Assuring everyone he’ll now face investigation back home, the US was presumably trying to avoid another embassy incident becoming a bilateral spat, like the 2019 death of the young local after a CIA spouse drove down the wrong side of the road. But this attempt at a low profile appears to be backfiring in spectacular fashion, with British tabloids already seeing another high-profile example of the US dissing British law.
🇷🇺 RUSSIA — Train attack.
A Russian drone has destroyed a train engine at the Ukraine-Polish border shortly after former British PM Boris Johnson, former CIA chief David Petraeus, and other Western security luminaries had departed a security conference in Kyiv. (BBC)
Comment: This seems like classic Putin salami-slicing: a direct hit on high-profile targets would risk a(nother) war he’ll lose, but this supposed near miss helps him intimidate Ukraine’s backers at lower cost.
🇸🇪 SWEDEN — Election drama.
Sweden’s Sunday election is a photo finish, with the centre-left opposition enjoying the slightest of leads over PM Kristersson’s pan-right coalition. Final results might not be out until Wednesday. (Al Jazeera)
Comment: A win for Sweden’s centre-left opposition might wobble the “resurgent right-populists” narrative across broader Europe, but seems unlikely to result in any major pivot in the way Sweden approaches (say) NATO, Ukraine, or even migration politics (caution is now mainstream).
🇹🇷 TURKEY — Crackdown on LGBTQ groups.
Turkish police spent the weekend raiding gay bars, NGO offices, and activist homes under an operation branded “My Family is Safe”. The justice minister is framing it all as part of President Erdogan’s wider ‘Decade of Family and Population’, aimed at reviving declining birth-rates. (DW)
Comment: The birthrate story is real, but pinging it on a small (and legal) minority feels familiar and political — define the nation as a tradition, anyone else as foreign, and thereby maybe distract from your own issues like rent and wages. The EU will either mutter something about rights or hold its nose to keep the Turkey file quiet. Watch which.
🇮🇳 INDIA — BRICS summit.
Narendra Modi’s big weekend BRICS summit has wrapped with a joint declaration hailing their “unprecedented unity”. (Al Jazeera)
Comment: Getting these odd bedfellows to agree on a 140-point declaration is solid, particularly given two (the UAE and Iran) are now on opposite ends of a shooting war. But beyond a shared loathing for US sanctions, words are where the consensus ends: Iran wanted condemnation of US strikes, but the Emiratis wanted condemnation of Iranian strikes, so they all channelled their Basil Fawlty and didn’t mention the war at all. Amid an India-China rivalry and Modi scrubbing anything that looks too anti-West, it still looks like a pretty performative grouping.
🇩🇿 ALGERIA — Cutting ties.
After years of tensions, Algiers has formally severed ties with the UAE and closed its airspace to Emirati-flagged aircraft. (EuroNews)
Comment: Algeria has long called out the UAE’s more interventionist foreign policy, whether backing a) rival Morocco’s claims over the Western Sahara, b) a rival faction in neighbouring Libya, c) the notorious RSF in Sudan’s civil war, or allegedly even d) Kabyle separatists within Algeria’s own borders. That last file is probably what broke the camel’s back, but the UAE’s relative weight advantage means Algeria now risks isolating itself.
Extra Intrigue
🤣 Your weekly round-up of the world’s lighter news
Evidence in Indonesia suggests humans have been getting high for 25,000 years.
Polish actors in a film about a treasure hunt have found an actual treasure.
European scientists have won a prize for using underpants to assess soil health.
A 1937 Springfield mural has gone viral for seemingly showing an Indigenous man holding a smartphone. Spoiler: it’s probably just a small mirror.
A Canadian man has entered the Guinness Book of Records for solving 211 Rubik's Cubes while bouncing on a pogo stick. Congrats?
From our sponsors
You're Invited: Tax-Smart Investing webinar on September 17. Learn the portfolio moves that can help you maximize your after-tax returns — join Range live, and bring your questions for Q&A.
Mystery of the day
The building hosting the (ex) Belarusian embassy and the South Korean embassy in Helsinki. Credit: Google Maps.
The charming slab of cement above contains none other than the Belarusian embassy in Helsinki. Or rather, it did until the mission’s closure earlier this month.
The Russian ally had already announced — without explanation — it’d temporarily suspend operations. Then, in true autocratic fashion, it made the closure permanent by simply repealing the original order that first created the embassy. Problem solved!
It’s a regional sport by this point: the dictator (Lukashenko) closed his Estonia embassy in 2024, and has been “optimising” his European footprint since 2021.
But why? When pressed, the Belarusian foreign minister eventually argued this latest embassy was simply “unnecessary”, which probably reflects a few drivers: Western sanctions on Putin’s junior partner have killed whatever trade and engagement was left; and embassies aren’t free for an economy smaller than Idaho.
Anyway, if you’re a Belarusian needing a new passport in Helsinki, your customer service desk just got upgraded to a ferry ride over to Tallinn then a drive to Latvia.
Today’s poll
Which theory do you think best explains AI's current 'slow down' moment?
Thursday’s poll: If you had $15B, where would you build more AI infrastructure?
🇺🇸 USA (18%)
🇨🇳 China (4%)
🇪🇺 EU (11%)
🇨🇦 Canada (47%)
🇮🇳 India (9%)
🇲🇦 Morocco (3%)
✍️ Other (Write in!) (8%)
Your two cents:
🇨🇦 R: “Canada wins for the cold weather, hydroelectric power, good powerline infrastructure, and stable government.”
🇲🇦 R.N.M: “Morocco has huge solar and wind resources and is close to Europe.”
✍️ T.S: “By no means would I invest in a soon-to-burst bubble.”
✍️ M.M: “Has no one seen the 'Terminator' movies? Let's spend $15 BILLION on renewable energy infrastructure instead.”
Honourable mentions: Australia, Argentina, South Korea, Nigeria, and India.
PS — shoutout to ‘RH’ in the group-chat for today’s Recursive self-importance email subject 😂. Become an Intriguer and join the banter today!