AI Can Now Design Functional Viruses. Should We Worry?(spectrum.ieee.org) |
AI Can Now Design Functional Viruses. Should We Worry?(spectrum.ieee.org) |
A lot of things could be an existential threat, nuclear weapons, bio-weapons, climate change, the end of the Flynn effect, cratering birth rates, whatever.
You can always pick out a particularly dystopian vision of the future if you're so inclined. The doom narrative is sells ad revenue, so it's a huge amount of what we all consume online, but it's not really that mapped out to reality...
Somethings get worse, some things get better, in general, the average QoL for the average human is getting better. If we decide not to go to global thermonuclear war, or we decide to unplug any killer AIs that we invent, we'll probably be ok. If we can reduce the population to 3,000-ish humans after the Toba eruption, we can probably survive whatever self-imposed calamities we can create.
I think there's something particularly noxious about the present nihilism trend I see in a lot of this doom themed content. People should go outside and touch grass.
So is nuclear annihilation. Yet, here we are.
> "I occasionally think how quickly our differences worldwide would vanish if we were facing an alien threat from outside this world." -- Ronald Reagan at the United Nations [0]
We are going to be fine as long as we remember to be nice to each other, starting towards the people who are adjacent to us.
Her conclusion to the latter book comes out and says it, paraphrasing here:
> Nuclear war is terrifying in its destruction, but the chain of command limits the risk of a rogue actor. Whereas bio weapons can be designed and manufactured in relatively low tech environments with equal or greater magnitudes of impact.
It'd be like releasing an invasive species into an ecosystem that has not co-evolved to co-exist with it.
I will kickstart a home DNA printer
I will name it Torment Nexus
I will be rich
[0] "43 freight containers and weighing in at 165 tons." - https://www.theregister.com/offbeat/2024/12/03/asml-releases...
[1] not in red blood cells
EUV is the pinnacle of a highly advanced supply chain, you could clone it but it requires many billions of dollars of R&D and massively specialized components; an economy like China likely can support the manufacturing tree required.
I would not be surprised if DNA printers seem end up closer to EUV or even more commoditized, if they are an extension of how all pharma/biotech manufacturing and research gets done.
But this is ultimately contingent on what the technology tree looks like. You couldn’t prevent diffusion of 3D printed ghost guns, even if you wanted to.
Versus EUV which one could dump literally unlimited amounts of money into solving and still probably fail for decades.
I will give you a classic example, the Lipid Nanoparticle used for the delivery of mRNA vaccine is the real lynchpin of covid vaccines. The media never talks about it. Any biology student knows how lipid bilayers work (they are the basic building block of cells, it's like asking a programmer what a terminal is). But the process of manufacturing those lipid nano particles at scale was the main bottleneck for delivering mRNA to the target cell. That was the trade secret. It's not sexy or hot like mRNA so the media doesn't talk about it. It's why many developing countries tried to make their own vaccines and failed completely. The mRNA was the easy part but what use is a bunch of mRNA if you can't get them delivered to cells?
Any biology student (even high school ones) can describe how a cell membrane works. But designing a man-made one that can be created at scale and be guaranteed to fuse with mammalian cells is mostly a matter of trial and error. It's a bit like finding viable pseudo linear memory use alternatives to the basic attention architecture in machine learning. Every lab has their own take on it.
With AI nowadays the search space of viable biotech ex nihilo experimentation is much smaller given that you can just prompt it to give you plausible protocols.
Designing a modifcation to an existing virus is perhaps more achievable, but the implementation is still the hard part.
Thermo Fisher sells a kit to do this [1]. It's something you might do a few thousand times on the way to actually generating a novel virus.
> I’d argue estimate the risks of a synthetically induced pandemic of massive impact this century are very high
I'd argue that it probably already happened, and had nothing to do with AI.
[1] AAV transfection.
But the next one (or the first one, if that one wasn’t manmade) will probably be much worse.
Although, I must admit, as a doomer, I'm pleasantly surprised things aren't already worse than they are.
Edit: I'm a PhD biochemist that has done a fair bit of virus modification. Love the downvotes. Never change HN. ¯\_(ツ)_/¯
Building a functional virus is not even remotely a turn-key operation. Assuming you could get the genome, expressing that genome in a way that results in infectious virus is non-trivial.
[1] Even then, the companies will rarely give you long enough reads to make a full genome scaffold. It looks like there are two companies that will do 50-100kb assemblies as a service today, but I guarantee you they're going to look at what you're trying to make, and who is asking them to do it.
They really just synthesize anything for you?
You may be skipping a few of the involved steps.
A profitable prediction markets trader who goes by the alias semiotic rivalry posted this prediction on his X account recently:
> Prediction: in 1-2 years, almost everyone will agree that open sourcing the best AIs is totally crazy as the AIs become capable of autonomous hacking of most humans and institutions, serious terrorism and creation of biological weapons, etc.
And I mostly agree with him. As much as it's awful to have such a concentration of capabilities in the hands of a few governments or corporations, as far as I can tell there are no easy answers here, only a matter of trade offs and different risks.
did you mean greater number of people with less resources?
EDIT: thanks to others that point out how you can interpret this differently, which I think is: "it is now easier for smaller teams with fewer resources" rather than my original interpretation of "power is now concentrated so that fewer people can do this than before"
Now, the worry is a lone nut in a basement with no money could pull off an act of bio terrorism.
If the Manhattan project could build a fat man in 4 years your state university's physics department has somebody who could reproduce that given the necessary resources. It might take them longer than four years given that they are not staffed with the modern equivalents of Oppenheimer, teller, et al but they could do it within a human lifetime if they actually wanted to cause a mass-casualty event and weren't afraid of the aforementioned JDAM.
However I think the result is disappointing, it concludes the only solution is a massive preventative surveillance state and global governance.
And technology doesn’t really care what you use it for. It can be constructive or destructive.
It “only” enhanced local destructive potential, until last century. Then it got to the point where a huge effort could produce a truly global catastrophe. And then a bit smaller effort. And technology continues to march on.
- can save my life,
- is interested in saving my life
- i have some control over
Yes, they are currently snatching kids of (illegal) immigrant parents and it is horrible, but who else would protect me against lone nuts in a basement, or robor armies of trillionairs?
I see it's still happening as of July 2026 -
https://www.techtimes.com/articles/319863/20260707/sunglasse...
https://www.reddit.com/r/TeslaLounge/comments/zt1kgt/anyone_...
But even though there are a few people tricking the "are you paying attention" check, probably more people complain about it being too careful and overbearing.
I feel like we are all somehow memoryhole-ing this massive trauma that nothing in this batshit era makes any sense without remembering.
Specifically, I wonder if the thirst for a machine god has anything to do with the activation of the Rapture narrative in the collective unconscious, which we then empower by forgetting its trigger. Call Joseph Campbell
* there's not an obvious benefit to being the one nation that says "you know what? We want to be known as 'the nation whose biolab insecurity wrecked the world', because taking blame works out so well for everyone.", though I'm fairly sure a large portion of the world will be totally willing to wargame the impact of hosting a lab as a deadman switch so nobody else attacks them first…
You email the RNA strand (a text file) to a lab along with some money. The lab converts that information into a physical object (read: virus) and sends it back to you in the mail.
What step is missing?
Distance from mailbox to live virus is a few days of bench work from a semi-competent biochemist.
And yes, this service claims to have safeguards. Does every service? Does every future analogous service? Are those safeguards guaranteed to be robust against totally novel pathogens?
Answers: No, no, and no.
You: "No, of course not. I once went on a website to buy a drug, and they made go to a doctor and get a prescription and send it to them"
The black market exists -- but comes with black market problems.
Still, even though your estimates are silly, they aren't silly enough to change the conclusion that this isn't a comforting security barrier, so I suppose we agree on that much.
https://en.wikipedia.org/wiki/Nuclear_holocaust#Likelihood_o...
Annie Jacobsen is awesome, her book about Area 51 was riveting, but I also cannot unsee "Capitalist Realism" and the idea that we keep picturing civilization-ending catastrophes because we don't want to go to work in the morning, and walking/biking everywhere and living in a small community is amazingly appealing to us at a base level. We yearn for the lifestyle of the apocalypse survivor, but none of us want to go through the apocalypse. Generally, I think we kind of say, "oh, it'll be the end of the world" because most of us (myself included) lack the imagination to really look beyond linear extrapolations of the status quo most of the time.
There you go!
Steps from your mailbox to an actual virus is a few days of bench work.
That's not a service that makes you a virus, and making one from an mRNA transcript is not trivial.
For each of the years below, what is YOUR estimate for how long would it take a person of 80th percentile intelligence with $10,000,000 (2026 money) to get from "I'd like to kill a lot of people" to a working virus?
You can provide a range for each if you'd like.
In the year 1800 – how long?
1950?
2020?
2026?
2050?
2200?
The point is that the rate-limiting step in producing something infective is a couple million dollars in special lab equipment, skilled practitioners, and incredibly tricky and tedious procedures that take time and repetition to master.
So yeah, you can try to pay someone who has that expertise to make your top-sekret superbug (good luck with that), or you can do what any real threat actor is going to do, and put in the time and investment. That's the hard part, not asking ChatGPT for a genome sequence.
Let me put it another way: despite the synthesis pathways for many common illegal drugs being widely known and trivial enough that an undergrad O-chem major could do them, in fact doing this is a very good way to get caught and arrested. And that's easy.
Glad we agree that there are labs that will solve very hard parts of creating a novel virus, either by accident or by negligence.
Here's a red team exercise where researchers successfully bought fragments of the best known pathogen in the world, never mind a novel one.
What reasonable precautions failed to catch this but we could expect to catch a novel virus that "kinda looks like" a known virus?
https://www.statnews.com/2024/05/08/shouldnt-be-easy-buy-syn...
I'm just saying that people are focused on the wrong problem because they've got LLM-brain.
The people "got LLM-brain" are saying: "Wow those barriers are coming down fast"
And you're saying: "There are still barriers"
Both are true, but only one is helpful for thinking about the future. The other is hardly even helpful for thinking about the present, and I've never seen a quasi-intelligent person do it in lieu of some other motivations in their reasoning.
Synthetic biology is one of those things.
LLMs are clearly going to solve a part that, 10 years ago, one would've assumed would be one of the highest and last barriers to go down.
Without LLMs, a reasonable security apparatus would be:
Step 1. Disallow dissemination and synthesis of known pathogens
Step 2. Watch closely anyone with the decades of training required to come up with their own pathogen
No.
No, the barriers I'm talking about aren't coming down. Not quickly anyway.
No, nobody assumed this was the "highest and last barrier". That's the point you keep missing.
LLMs are not "solving" the rate-limiting step here. It only looks that way if you don't know what's involved.