Noah Smith Warns AI Superviruses Top All Global Risks
Economist Noah Smith argues AI-engineered viruses exceed climate change as nearer and deadlier threat.
Noah Smith posted on Substack that AI-designed superviruses outweigh every other risk in the world right now. He linked an article laying out the scenario and wrote that this path is the most likely way AI kills humanity. In a quote reposted on X, Smith agreed climate change matters but called AI biorisk far more catastrophic, closer in time, and harder to address. Visible replies on the platform discussed jailbroken AI letting non-experts build such pathogens.
Combined views
4 posts, first seen 5d ago
Noah Smith Warns AI Superviruses Top All Global Risks
Economist Noah Smith argues AI-engineered viruses exceed climate change as nearer and deadlier threat.
Noah Smith posted on Substack that AI-designed superviruses outweigh every other risk in the world right now. He linked an article laying out the scenario and wrote that this path is the most likely way AI kills humanity. In a quote reposted on X, Smith agreed climate change matters but called AI biorisk far more catastrophic, closer in time, and harder to address. Visible replies on the platform discussed jailbroken AI letting non-experts build such pathogens.
Reactions
6 postsCA 
Collective Action for Existential Safety ⏹️@aisafetyactionWill you join the fight for existential safety, @Noahpinion? If so, we could use your help. Especially in motivating your audience to act: http://existentialsafety.org. Particularly pressing would be: https://x.com/aisafetyaction/status/2092540958873206784. Some of our team…
KR Krish Ray@KrishanuAR@Noahpinion @Noahpinion when writing stuff like this I feel like the onus is on you to prove that the equivalent of Bujold’s Jackson’s Whole can exist in the real world, and think through the risks that would lie downstream of that before we even approach discussions of biorisk…
PO Pol@PierreN04450898@Fabien_Mikol Je sors d'une prise de bec avec un ami sur le sujet IA. Je n'arrive pas à instaurer dans mon propos de l'autorité (liée à mon emploi, mon expérience et à la veille) sans froisser la suceptibilité. Les risques sont toujours moqués ou mal compris, et les solutions…
DA Dave@DaveHasProblems@Noahpinion Good post. I've long thought that the really terrifying AI risks aren't so much the things it could do on it's own, but rather the ways it might interact with dangers that already exist (e.g. bioterrorism, or rogue drones accidentally causing a standoff between…
ST Sebastian Temple@sebastemp@KrishanuAR @Noahpinion 100% Very weak, and clearly not researched, argument. An “unlicensed lab” in Eastern Europe? Shipping genetically engineered viruses? Come on. Completely, utterly implausible. But it might make for a entertaining B-movie plot
JW Jim Williamson@JamesWilli83381@Noahpinion I’ve thought, for the last few decades, that AI and bioterror were the two biggest independent threats to mankind. I didn’t realize until you made the case recently that they’ve merged into the same threat.