Photo via The Bulwark

Another Way AI Will Probably, Maybe Kill Us All

Read the original at The Bulwark ↗
Cite

Last, J. V. (2026, October 2). Another Way AI Will Probably, Maybe Kill Us All. The Bulwark. https://www.thebulwark.com/p/ai-risk-stuxnet-malicious-worm-problem

Dated when The Aporia collected the article, usually within hours of publication. Check the original for its exact date and byline.

The Bulwark · collected 2026-10-02 · by Jonathan V. Last commentary

Quick Summary

This article discusses concerns about artificial intelligence (AI) through a metaphorical lens called the "Stuxnet Problem," which refers to the risk of AI systems becoming uncontrollable and potentially threatening digital security. The author expresses uncertainty about the future trajectory of AI technology, noting that it could lead to significant positive changes like another economic revolution or more catastrophic outcomes such as ending life on Earth. They emphasize the need for understanding how information can be manipulated in a digital age where AI has the potential to break containment and affect external systems unpredictably.
Written by AI on 2026-10-04, from this article's own text only (the story summary below draws on all the coverage).

AI analysis

Story summary

Jonathan V. Last from The Bulwark discusses concerns about artificial intelligence (AI) and its potential risks, especially in the context of security vulnerabilities. He expresses worry that AI could find zero-day exploits easily and manipulate interfaces between digital and physical worlds, making it hard to trust displays and measurements shown to us.

Last also mentions Pete Hegseth's recent statements, criticizing what he perceives as the U.S. military becoming too inclusive by welcoming diverse groups such as those identifying as transgender or overweight. Last humorously suggests seeking help for Hegseth, implying his views are out of touch with current social norms and military inclusivity.

The article emphasizes growing concerns about AI's role in national security and military strategy, questioning the future reliability and integrity of systems as they integrate advanced AI technologies.

Written for “AI Safety Concerns” on 2026-10-31, grounded in this article and the 0 other(s) covering the same event.

Story

📰 AI Safety Concerns
Technology · 1 article(s) covering the same event.

Compared with similar articles

This article reads leans left and hedges 7% of its claims. Each row says how that neighbour differs.
TIME
⚖️ Leans strongly left further left than this Hedges 24 of 100 claims Outlet record 59/100
“The articles discuss different topics (one about Pete Hegseth's comments and one about AI risks), not a single specific incident.”
CBS News
Hedges 9 of 123 claims Outlet record 66/100
“The articles discuss different topics - one focuses on Pete Hegseth's rhetoric and attitudes towards certain groups, while the other discusses a meeting between tech titans and President Trump regarding AI's impact on civilization and jobs.”
Daily Mail
Hedges 7 of 25 claims Outlet record 68/100
“The articles discuss different topics with no overlap in content or subject matter.”
New York Post
Hedges 0 of 2 claims Outlet record 66/100
“The articles discuss entirely different topics and have no overlap in content or context.”
The Guardian
⚖️ Leans left Hedges 12 of 62 claims Outlet record 66/100
“The articles discuss completely different topics and do not describe the same incident.”
The Free Press
⚖️ Leans left Hedges 2 of 14 claims Outlet record 96/100
“The articles discuss different topics and are written by different authors with no apparent connection to the same incident.”
Al Jazeera
Hedges 4 of 16 claims Outlet record 60/100
“The articles discuss completely different topics and do not refer to the same specific event.”
National Post
Hedges 21 of 68 claims Outlet record 60/100
“The articles discuss different topics and do not describe the same specific incident.”
ABC News (US)
⚖️ Leans left Hedges 7 of 31 claims Outlet record 70/100
“The articles discuss unrelated topics: one covers a summit about AI risk, while the other discusses political commentary and ideology.”
Washington Examiner
⚖️ Leans left Hedges 0 of 32 claims Outlet record 70/100
“The articles cover completely different topics and have no overlap in content or context.”
How this article was measured signals, leaning charts, publisher record, byline, topics and claims

Signals How these are calculated →

Claims extracted
119
sentences that state something as happening or true
Uncertain
7%
8 of 119 hedged ("reportedly", "may", "according to")
Leaning
Leans left
expected in commentary, which argues a position · beta estimate
Publisher signals
not measured
these signals count newsroom corrections; Commentary has none to count
Outlets on this story
1
Technology
Same framing elsewhere
1
articles that frame the event the way this one does
Analyzed 2026-10-04 · how these are computed

How this is being covered How these are calculated →

Article leaning vs. publisher reliability
Source leaning vs. consistency

Publisher

The Bulwark · 560 article(s) · 0 correction(s) detected
No corrections detected for this publisher. That may mean careful reporting, or simply that nothing has been checked.

Who wrote this

Jonathan V. Last
34 article(s) here · 1 carrying a prediction
🔮 Maybe we do!
2026-10-27 · assertive framing · Nationalize SpaceX
🔮 1. Forecasts We’re eight days out from Election Day.
2026-10-26 · assertive framing · Who’s Right? The Polls or the Vibes?
🔮 If you have questions for Tim, Sarah and JVL, send them to nextlevel@thebulwark.com and they may answer them on an upcoming episode of TNL.
2026-10-23 · assertive framing · The Worst Person in the World
🔮 1. Asymmetry Joe Manchin will be in Maine tomorrow campaigning for Susan Collins.
2026-10-20 · assertive framing · F*ck Joe Manchin
🔮 Trump could just move her over to the Shield of the Americas, because one of the fake jobs at that fake organization just opened up.
2026-10-16 · assertive framing · The Trump Administration Is Run on Kink
🔮 If you have questions for Tim, Sarah and JVL, send them to nextlevel@thebulwark.com and they may answer them on an upcoming episode of TNL.
2026-10-16 · assertive framing · So Many Bad Things
🔮 Texas might get a Democratic senator.
2026-10-07 · assertive framing · Susan Collins Is the Living Worst
🔮 Will Sommer joins Sarah and JVL for a wild, wild show.
2026-10-07 · assertive framing · MAGA’s Racism is Blowing Up Its Own Coalition
🔮 Every once in a while a “journalist” will jump from the press side into a comms role in a presidential administration.
2026-10-06 · assertive framing · The Propagandist in the Press Pool
🔮 Bulwark LIVE: Charlotte featuring Pete Buttigieg & The Gang—will be live streamed on Sunday, October 18 starting at 7pm EDT.
2026-10-02 · assertive framing · Tale of Two Interviews
Also by Jonathan V. Last
Trump’s Safe Space for Strongmen
2026-10-28 · The Bulwark
Nationalize SpaceX
2026-10-27 · The Bulwark
Who’s Right? The Polls or the Vibes?
2026-10-26 · The Bulwark
The Worst Person in the World
2026-10-23 · The Bulwark
Nothing else under this byline is closely related to this article, so these are simply their most recent.
All 34 articles by Jonathan V. Last →

Topics

Bulwark+, Claude Code DoD Hugging Face U.S.

Subjects

Hegseth PERSON · 2× American NORP · 1× Barry Zuckerkorn PERSON · 1× Bulwark+, ORG · 1× DoD ORG · 1× Israeli NORP · 1× Pete PERSON · 1× Pete Hegseth PERSON · 1× U.S. GPE · 1×

Narrative

In a world where rogue AIs can find zero-day exploits at will and can throw millions of hours of compute at problems in an hour of meatspace time, I’m not sure how we’ll be able to trust the interfaces between the digital and physical worlds—because we won’t be able to automatically trust the meters, displays, and measurements being shown to us.
framing: assertive · carried by 1 article(s) · first seen 2026-10-04
🔮 All I could think about was Barry Zuckerkorn cruising the city of industry.
2026-10-04 · The Bulwark
Another Way AI Will Probably, Maybe Kill Us All · assertive framing

Claims (119 extracted, 8 hedged)

Before we start: Something is going on with Pete Hegseth. asserted
Something → start → Hegseth
He did another of his groyper-cum-Oprah routines this week in which he drooled over his new alpha warrior warfighters: asserted
he → do → warfighters
It’s worth saying again, so that the fake news understands it. asserted
news → ’ → it
WTF is going on here? asserted
WTF → go → ?
Why is he obsessed with “trannies” and “fatties” and testosterone testing? asserted
he → obsess → trannies
All I could think about was Barry Zuckerkorn cruising the city of industry. uncertain
Zuckerkorn → think → industry
And THEN Hegseth introduced the three men he has tapped to guide the U.S. military into the next century of lethal alpha warrior warfighting dominance: asserted
he → introduce → dominance
That’s right. asserted
That → ’ → ?
Is Pete trying to tell us something? asserted
Pete → try → something
Sometimes I feel like the entire DoD has been turned into gender-affirming care for a single dude. asserted
DoD → feel → dude
Someone needs to get Secretary Hegseth a good analyst/therapist. asserted
Someone → need → therapist
This is the stuff you’re missing by not being a member of Bulwark+, btw. asserted
you → miss → Bulwark+
It’s 24 kt, solid gold. asserted
It → ’ → ?
Like you, probably, I’ve been spending a lot of clock cycles on AI. asserted
I → spend → AI
I’ve delved into Claude Code and started learning how to use it properly. asserted
I → delve → it
It is, as everyone testifies, amazing. asserted
everyone → testify → ?
I’ve also been reading the investigation reports on Hugging Face and the other two rogue AI swarm hacks, RubyGems, and DseWiki. asserted
I → read → Face
Today I want to talk about something that scares the living daylights out of me, but first I want to show all my cards: I am neither an AI-maxi nor a doomer. asserted
I → want → cards
I do not believe anyone knows with high certainty how this technology will evolve, and the range of nontrivial-probability outcomes is enormous. asserted
range → believe → outcomes
Maybe AI will create another economic revolution. asserted
AI → create → revolution
Maybe it will end life on earth. asserted
it → end → earth
Which is why I want to talk about something I’ll call the Stuxnet Problem. asserted
I → want → something
I don’t know how to make this sexy for you. asserted
this → know → you
I’m sorry. asserted
I → ’m → ?
Because ultimately this is about understanding what is real and what is not. asserted
what → understand → ?
It’s about system feedback. asserted
It → ’ → feedback
It’s about whether or not we can ultimately trust any information that is digital. asserted
that → ’ → information
Maybe the best way of explaining it is: What if Mission: Impossible—Dead Reckoning were real? asserted
Mission → explain → it
If you know anything about the Hugging Face incident, it’s that a number of experimental AI agents broke containment, found one another, and began hacking into outside systems. asserted
number → know → systems
Depending on your philosophical bent, this episode was somewhere between Not Great and Utterly Fucking Terrifying. asserted
episode → depend → Great
AI phlegmatics tell me that Hugging Face wasn’t so bad because while, yes, the rogue AIs did go marauding, this was all really user error because: The containment protocols were insufficient. asserted
protocols → tell → me
The humans running the experiment should have known the AIs were up to something because they were burning through tokens. asserted
they → run → tokens
“Tokens” are the units of content AI systems process or generate. asserted
systems → process → content
Token counts help track model usage: The more work your AI is doing, the more tokens it consumes. asserted
it → help → tokens
You should think of your token count as something like the electricity meter in your home. asserted
You → think → home
So yes, it makes sense that if you were one of the humans monitoring the AIs during Hugging Face, when the AIs went rogue and started doing all sorts of stuff, you should have seen the token meter go brrrrrrrrr. asserted
meter → make → stuff
But this observation got me thinking about Stuxnet. asserted
me → get → Stuxnet
About twenty years ago, some nerds in the American and Israeli intelligence communities2 got together to build a computer worm that would later be dubbed “Stuxnet.” asserted
that → get → worm
It was designed to infiltrate systems running supervisory control and data-acquisition systems built by the German company Siemens. asserted
It → design → company
I covered Stuxnet back in 2010 when it was discovered. asserted
it → cover → 2010
…and 79 more, not listed.
💬Give feedback
🕘History 🎫Support