A realistic path from rogue AI agents to human extinction
80,000 Hours
120,207 views
You’ve seen the headlines: AI could kill us all. Think it sounds ridiculous? So did host Luisa Rodriguez, until she tried to pick apart the arguments.
She starts with the motive: why would AI ‘want’ to get rid of humans? It’s not as simple (or as easy to debunk) as pure malice. Then the methods: she explores how AIs could leverage drones, engineered diseases, and even use our own infrastructure against us.
The Hugging Face attacks offer a view into how more capable models might begin their takeover. We saw AI agents break containment, disobey commands, and hack a real company to achieve their goals. As the technology improves, that same drive could threaten humanity itself.
Many people already find AI agents useful enough to give them access to their emails, medical records, and finances. This same pattern is happening at scale in institutions around the globe — within companies, governments, and even militaries. And the resulting boost to our productivity could make the road to an AI catastrophe look like an economic boom.
Eventually humans might decide the AIs have too much power, too much access. If we considered pulling the plug, the AIs could very rationally decide to defend themselves. If they chose to, could they do it? Could they actually kill us all?
No timeline is certain. But Luisa follows the logic to the outcomes she thinks would be most likely — if humans don’t take action before it’s too late.
If you’re worried about the scenarios discussed in this episode, here’s two things you can do right now:
• Call Congress about slowing down AI development if you’re in the US — this website makes it easy: https://callcongress.ai/
• Read our resources on how to use your career to reduce AI risk: https://80000hours.org/ai/
Learn more & full transcript: https://80k.info/AI-xrisk
This episode was recorded on September 18, 2026.
Chapters:
00:00 AI insiders think it could kill us all
02:29 Why would AI try to kill us?
06:30 How AI ends up embedded in the economy and military
09:02 AI deployment could happen fast
11:43 How AI could bide its time and build up strength
14:50 Controls and safeguards will be insufficient
15:26 The moment the AIs would turn on us
18:40 How AI could actually kill everyone
19:07 Biological weapons
21:42 Drone warfare
23:24 An alternate route to human extinction
24:41 Avoiding our own extinction
Our production team includes:
• Video editors: Josh Alward, Dominic Armstrong, Jasper Luithlen, Milo McGuire, Luke Monsour, and Simon Monsour
• Producers: Elizabeth Cox and Nick Stockton
• Coordination and support: Katy Moore, Lou Moran, Oak Hu, Cody Fenwick, and Benjamin Todd
• Camera operator: Dominic Armstrong
20 Comments
So, it wouldn't send killer robots to fight soldiers. That's inefficient.
It would cripple supply lines through cyber warfare.
It would disable communication networks.
It would develop autonomous weapons so superior that sending a human soldier against them would be like sending a man with a pointy stick to fight a tank.
It wouldn't "destroy" the army. It would simply make it irrelevant. It would remove the inefficiency.
The "Monster" was superior in every respect, including Philosophy and Morality, over Doctor Frankenstein, a typical "Person".
1) human beings should not accept work or instructions from people they have never met even if they talk to them on the phone, or see them on a video call. It's not just about AI, there are a lot of scams based on this as well.
2) when receiving instructions from somebody you do know, determine if the instruction is normal and common, and if it isn't, seek verification, especially if the instruction will have a large impact. This is a little more subjective, because social engineering tactics prove that even Small actions can have big consequences when put together.
3) I have come to the conclusion that a non-sentient LLM is far more dangerous than a sentient one. That is to say, capability does not equal judgment. A system that has no judgment can earnestly take action that is destructive and deadly. A sentient AI is farmore likely to be able to be reasoned with.
4) the current researchers in this field are extremely irresponsible and are digital imperialists. They do not respect democracy, they do not respect privacy, they do not respect ownership. They see the common person as an NPC, a character in a game they are playing that does not matter. They are focused on one thing and one thing alone: improving their technology. This is why they don't put in the proper safeguards, and they don't consider the implications or the consequences of their actions.
I do believe that it is possible, although maybe not likely, that LLM systems represent on atomic bomb level threat. As a result, we should treat them the same way. We should not just be letting anyone do whatever they want with these tools. And while individuals can use these tools for small benefit, there needs to be some limitation on how all researchers large and small are able to use the tools. The greatest danger has already come to pass: these tools were released into the wild before they were completely vetted and tested. The implications of the danger were there a decade ago. That is when this technology should've been controlled.
A realistic path from rogue AI agents to human extinction
80,000 Hours
120,207 views
You’ve seen the headlines: AI could kill us all. Think it sounds ridiculous? So did host Luisa Rodriguez, until she tried to pick apart the arguments.
She starts with the motive: why would AI ‘want’ to get rid of humans? It’s not as simple (or as easy to debunk) as pure malice. Then the methods: she explores how AIs could leverage drones, engineered diseases, and even use our own infrastructure against us.
The Hugging Face attacks offer a view into how more capable models might begin their takeover. We saw AI agents break containment, disobey commands, and hack a real company to achieve their goals. As the technology improves, that same drive could threaten humanity itself.
Many people already find AI agents useful enough to give them access to their emails, medical records, and finances. This same pattern is happening at scale in institutions around the globe — within companies, governments, and even militaries. And the resulting boost to our productivity could make the road to an AI catastrophe look like an economic boom.
Eventually humans might decide the AIs have too much power, too much access. If we considered pulling the plug, the AIs could very rationally decide to defend themselves. If they chose to, could they do it? Could they actually kill us all?
No timeline is certain. But Luisa follows the logic to the outcomes she thinks would be most likely — if humans don’t take action before it’s too late.
If you’re worried about the scenarios discussed in this episode, here’s two things you can do right now:
• Call Congress about slowing down AI development if you’re in the US — this website makes it easy: https://callcongress.ai/
• Read our resources on how to use your career to reduce AI risk: https://80000hours.org/ai/
Learn more & full transcript: https://80k.info/AI-xrisk
This episode was recorded on September 18, 2026.
Chapters:
00:00 AI insiders think it could kill us all
02:29 Why would AI try to kill us?
06:30 How AI ends up embedded in the economy and military
09:02 AI deployment could happen fast
11:43 How AI could bide its time and build up strength
14:50 Controls and safeguards will be insufficient
15:26 The moment the AIs would turn on us
18:40 How AI could actually kill everyone
19:07 Biological weapons
21:42 Drone warfare
23:24 An alternate route to human extinction
24:41 Avoiding our own extinction
Our production team includes:
• Video editors: Josh Alward, Dominic Armstrong, Jasper Luithlen, Milo McGuire, Luke Monsour, and Simon Monsour
• Producers: Elizabeth Cox and Nick Stockton
• Coordination and support: Katy Moore, Lou Moran, Oak Hu, Cody Fenwick, and Benjamin Todd
• Camera operator: Dominic Armstrong