Subscribe
Share
Share
Embed
Can biotechnology be used in malevolent ways? The simple and inexpensive ways to generate genetic material make creating viruses easier than ever. Dr Kevin Esvelt from MIT's Media Lab discusses how DNA could be used to drive a new pandemic, or become the basis of biological weapons. The discussion addresses the realistic potential malevolent use of biological tools, the risk of molecular biology tools being weaponized, along with potential mitigation strategies.
Talking Biotech is a weekly podcast that uncovers the stories, ideas and research of people at the frontier of biology and engineering.
Each episode explores how science and technology will transform agriculture, protect the environment, and feed 10 billion people by 2050.
Interviews are led by Dr. Kevin Folta, a professor of molecular biology and genomics.
The Risks of Manufactured Viruses - Dr. Kevin Esvelt, MIT Media Lab, MIT
Talking Biotech Podcast Episode 379
===
[00:00:00] Kevin Folta: Hi everybody, and welcome to this Week's Talking Biotech podcast by Colabra. Now, every week we get together to talk about breakthroughs in biotechnology. Lots of good things that are happening in agriculture or medicine or conservation. There's a lot of really positive, exciting things that happen, but for every yin, there's a yang.
And maybe it's time to let a little air out of the biotech balloon. I mean, there's all these good things that can be done, but can we ever find examples where this technology may be used for malevolent purposes? And I was very fortunate to be part of a meeting a few weeks ago. That discussed just that what are some ways that synthetic biology is being used and how may it be used by individuals who maybe want to use it in ways that wasn't supposed to be used?
And I got to meet an expert on this particular topic. So today we're speaking with Dr. Kevin Feld. He's an associate professor in the media arts and sciences at M I T, in the M I T Media Lab. Well, welcome to the podcast. Pleasure to be here. Yeah. This is really cool. or terrifying . So let's start at the beginning and it's, a lot of people maybe don't realize how easy it is to create a genetic sequence of interest.
So how easy is it to synthesize d n a and nowadays? How practical is it and, and what does it cost to do it?
[00:01:30] Kevin Esvelt: Well, it depends on how much DNA n a you want. . If you want a single gene, less than a thousand base pairs, you probably just want to order it as a single chunk of DNA, and then you can put it wherever you want.
That'll cost you on the order of 10 cents a base, something like something less than that. So a thousand base pairs gives you about a hundred dollars.
[00:01:57] Kevin Folta: So, so not, not too bad. And, and these days you could synthesize, you know, a large chunk of a gene of interest or, you know, could you do something as bold as to see as to produce an entire virus?
[00:02:11] Kevin Esvelt: Well, if you can make a whole metabolic pathway, and we certainly do that all the time, then yeah, you can make a virus and folks do
[00:02:20] Kevin Folta: really, so what, what , do you have any good examples of that? I mean, you say folks do, was this a clandestine project that was busted by the F B I or you know?
[00:02:30] Kevin Esvelt: No. No. So there's a whole branch of science that's based on tinkering with what's already.
Making changes and often breaking it and then see how it behaves. So you wanna know what a gene does? Usually easiest way to find out is you break it or you tweak it. So you make twice as much of the relevant protein. W for viruses, very often the easiest way to do this is to convert the virus back to a form made out of DNA n a in case it's made of RNA or something like that, and then make the changes you want in the dna N.
and then boot up the original virus from your DNA N in a cell. Then you can see how the virus behaves. Now that it, now that it's been tweaked in the ways that you want it, but once DNA synthesis got cheap enough, that means for lots of viruses, it's actually easier to just order the new viral genome from scratch.
And it turns out that DNA's D n A, it doesn't matter whether you got it from a virus. Or whether you synthesized it yourself, but Eckhard Wims group made it, it was a polio virus, and that was for research purposes, of course, just to show that you can now make a virus. And they did some really important work on changing the sequence of the virus to make it just a little bit less effective at making its proteins with every change.
And this is really important for developing vaccines because poliovirus vaccine, Has this nasty flaw where it can revert the changes in the vaccine to become the active nasty virus again. And in fact, we've eradicated wild type polio from the world. So the only kinds of polio that's left is the vaccine revertant viruses.
But if you use this new trick that they developed, Then there's, it's like death by a thousand cuts. You're crippling the virus, but there's no one single change To revert instead would have to revert dozens of different changes in order to become actually harmful again. So really great intentions, really great outcomes, but means we can now make viruses from scratch.
[00:04:42] Kevin Folta: Well, let's drill down on that a little bit more. You know, so many viruses are RNA based. So you're saying that you could actually order this DNA N and somehow get this DNA n into a cell and then have that DNA n be well in a laboratory of course. And then have some sort of a biological tissue culture or whatever be able to make a virus.
Because once you make the r n a, you now have the message to create the self-assembly machinery to make more of that virus.
[00:05:12] Kevin Esvelt: That's exactly right. So it depends on the kind of virus. Some RNA viruses need their proteins to be available when they get put in the cell. So then you need to make some D N A that will make the critical viral proteins.
And then you make the piece that will turn into the viral genome and you put 'em all in the cell together, and then the proteins start replicating the genome and boots. the life cycle of the virus. Now mind you, my lab mostly works with viruses that infect bacteria, but we also have moved into doing some work in viruses that infect mammalian cells, particularly mouse viruses.
And so that's how I can give you a fairly recent update on how hard this really is today, because I've seen a couple of my students take a stab. Folks who hadn't done serious virology before, but they're general bio technologists. And for things like influenza virus, it's fairly straightforward. Boot on the first try.
For things like coronavirus, it's much bigger than influenza, and you need to stitch together the whole viral genome. And even if you're a specialist in stitching together pieces of dna. N not all that easy, still doable, but a lot more d. .
[00:06:29] Kevin Folta: But even something like say influenza virus. I don't know how big that genome is.
I know coronavirus like 30 KB or something. Yeah, that's right. But, but we, yeah, but influenza virus is
[00:06:39] Kevin Esvelt: about how big, so it's about 15 ish, but the thing about it is it's segmented. It's an eight different pieces. And that's right. When you order synthetic dna, usually if you, you can either order it as a single chunk, you can add it cloned into a plasmid for you already ready to be expressed.
and in the case of influenza, you need, it's one of those negative sense ones, so you need the viral proteins to be booted up. There's a standard set of plasmas you can get for that. And then for the influenza strain of interest, you need to need the genome of each segment, and then you order all eight segments individually and they're all under 3000 base pairs.
So your total cost given, you know, bulk order and so forth, and if you order at one of the decently priced places is you can probably get under a thousand dollars.
[00:07:31] Kevin Folta: Yeah, and Al-Qaeda made that at their fudge sale last week. I mean, you can imagine that this is a pretty scary thing because if you can start to develop viruses from custom dna, Has there been any?
I It is, and this is just mind boggling. I'm a million questions. It's, it's one of the few times I've ever been speechless. How easy would it be to take a non pathogenic virus that, that we're infecting us all the time and actually make it pathogenic or weaponize it?
[00:08:02] Kevin Esvelt: Kevin, you don't sound very speechless.
I gotta say. Now, as to your question. Supposing it were really super easy. Would you really want me to tell you on the podcast ? But the thing is, the way norms in science go in bylaw, we're interested in fighting nature, right? And nature doesn't take the things we've learned and use them against us. So we're not really used to thinking about stuff that way.
And that means that the norms in the life sciences were pretty heavily. , you discover something wrong, you share it with the world so we can fix it.
And that's, shall we say, is not how they do it in fields where they're more used to human attackers. And so right now, I'm happy to say that there really aren't any obvious candidates for viruses that are definitely gonna take off and cause new pandemics. If someone were to synthesize and. The problem is it's really hard to imagine a future in which we don't learn how to make some pretty nasty things like that.
And in fact, there are a number of ongoing research programs that are deliberately trying to find sequence, characterize, and publicly advertise the existence of new pandemic capable virus. And this is done with the absolute best intentions under the reasoning of, you know, what nature might hit us with next.
We can get a headstart on developing defenses, maybe even prevent the natural virus from spilling over into humans in the first place. But the flip side is you sequence the virus, you run the key experiments that tell you, is this thing likely pandemic capable? And then you tell the world about it. Well, there's a lot of protocols.
For reverse genetics in virology that tell you how to boot up viruses of different families. And again, for influenza, it's pretty easy. You don't need to stitch the d n A together yourself. You can just order it comes in the mail, throw it into the cells, and it boots up because influenza's super well studied.
That doesn't mean it's easy for other viruses, but what we're talking about here, Order the DNA and follow the detailed step-by-step protocol because the scientific community will probably have done all the rest of the work of figuring out which viruses are dangerous and with the best of intentions, told the world what they are.
In fact, in a ranked order list, You can go to spillover.global, if you wanna see what I mean. ,
[00:10:46] Kevin Folta: but what about ones that we understand pretty well already? What about viruses like smallpox where, I don't know if it's in GenBank or not, but could you? Oh, it's, it's, they're all
[00:10:56] Kevin Esvelt: in, they're all in GenBank.
Every last one of them is in GenBank. But happy thing about smallpox, it's huge. It's 186,000 base pairs and it requires another smallpox to boot up that. , it relies normally on having a bunch of proteins go into the cell with it. And so you need those proteins in order to replicate the genome and get the life cycle started.
Even though it's a DNA virus, you can't just put the DNA n a in and have it work. And that means you not only need to make the whole smallpox genome, but you also need another pox virus. So it's a huge assembly project, and then it's a complicated booting up procedure. So, but this is why it was a fuss.
They made horse pox, which is a closely related version because that protocol could be used by a sufficiently skilled lab to make smallpox, but it wouldn't be easy. That is, there's probably only a hundred ish labs in the world that could do it. Now, mind you, smallpox killed half a billion people in the century.
Before it was erad. But on the flip side, it was eradicated. We know how to handle it, and the US has 300 million doses of vaccine stockpiled, so it's actually not one of the ones I worry. .
[00:12:10] Kevin Folta: So what are the safeguards that are built into the current way in which DNA n a is ordered or or designed through the companies that make it?
[00:12:18] Kevin Esvelt: So this was flagged by a number of farsighted folks way back in 2007, and remarkably, even though it costs the synthesis companies money to screen orders and make sure you're not trying to order a nasty virus or something like a toxin ricin, et cetera, , they got together and formed what's called the International Gene Synthesis Consortium, which is a bunch of firms that basically say, you can only join us if you can show that you actually screen your orders.
But it's voluntary, and in fact, it's a purely voluntary process. There's no legal requirement that they do it. But the biggest firms all joined and they continue to do it, and they claim that about 80% of synthetic d n A gets. . So in one sense, this is super encouraging, right? Here's a bunch of folks in industry doing the right thing, even though it costs them money.
This is how we wanna see the system work. So kudos to them. They're keeping us all safer. But there's just one little problem, which is that if you go to the I G S C website, you can see all of the companies that screen. If a company screens and doesn't join the I G S C, well, why is that? Obviously you wanna advertise that you're one of the responsible and safe ones, right?
So if you want an order that isn't screened, then you just need to find a provider that isn't on the list and order it from them. So right now there really aren't particularly good safeguards.
[00:13:55] Kevin Folta: Well, this is really interesting. Well, because you don't even need a company to do it. I mean, can't you just buy a, a DNA synthesizer on the secondhand market these days and kidnap a nerd and, you know, and, and create DNA
[00:14:09] Kevin Esvelt: Oh, I'm, well, I am a nerd, but I actually did my PhD in a chemical biology lab. So I guess you can say I'm one of those rare people who both knows how to work a current, at least phosphoramidite based chemical DNA synthesizer. and can assemble the resulting small pieces of DNA into larger things up to viral genome size, and could probably boot up at least a simple virus like influenza.
But most of the people who can boot up the virus don't have the skills to work the oligo synthesizer or the assembly. So you're down to a much smaller group of people who can do that on their own. But to your point about kidnap a nerd, and if you can make them do it or find some nerd who's comparably nasty purposes, yeah, that's pretty.
[00:14:53] Kevin Folta: Yeah. It seems like you could train up on this and figure it out. I mean, it, the instruction manual's gotta be pretty dense, but it seems like it's the kind of thing where you could possibly be able to like a, a, a. Malicious actor or y you know, group could potentially buy a DNA n a synthesizer and figure it out.
I mean, if you could figure out how to fly a plane into a building, this seems like a, a much easier step to accomplish.
[00:15:19] Kevin Esvelt: You know, you could, but the lucky thing is terrorists throughout history have been remarkably incompetent.
[00:15:25] Kevin Folta: So, so you don't, you don't necessarily think that this is something that is on a terrorist menu is to, you know, things to do.
[00:15:35] Kevin Esvelt: I don't think it'd be easy for them. I guess I'm most worried about a different caliber of terrorists from the, so from the folks that we historically have worried about that is to say, I mean, first of all, yeah, you know, there's definitely some Al-Qaeda types who were. Given the opportunity to cooperate in ways that wouldn't harm their interests and would protect women and children on both sides basically said, no, we'll do anything to kill you and yours, no matter what the damage it takes on innocence.
That's just how it is. So would folks like that deliberately make pandemics? Possibly. But I guess I'm probably more worried about someone like, you know, the uni bomber. Because here's a guy who in the early 1980s wrote about the immense power of biotechnology. Harvard undergrad, Berkeley mathematics professor who murdered a bunch of people because he wanted to bring down the industrial system that was destroying human dignity to condense his 30 thou 5,000 page manifesto into a single line.
But is this the kind of guy competent? To learn the relevant skills if you had to. Yeah, probably so. I guess it's a question of how long do you think it will take once there is an archive with credible blueprints for plagues, how long's it gonna be before someone deliberately mix and release? and I've asked a bunch of people this question.
You get answers that are all over the map and it varies tremendously by how easy you think it is to make the plague, of course. And thing is something like influenza, again, it's fairly straightforward. You don't have to do molecular cloning, but that's still not to say that it's easy, it's. , if you calculate how many people have gotten PhDs in virology, presumably, many at least of which can do it.
And my PhD certainly isn't in virology. I'm, I'm a biochemist technically, but I learned the relevant skills. Folks in my lab aren't getting PhDs in virology, and several of them can do it. So when you crunch the numbers just on the virology, PhDs, you come to about 30,000 people world. In the last 20.
[00:18:04] Kevin Folta: You know, it's about 30,000 too many
Well,
[00:18:08] Kevin Esvelt: that's the thing I don't really understand when peop, you know, when I relate this, a lot of folks sometimes say, well, I shouldn't say a lot of folks. Actually, it depends. A great deal on your background, whether you're from science or engineering. Engineers. Tend to respond as you have. Some of the scientists are like, ah, well, it's actually not that easy.
It's harder than you're making it say. A lot of them think I'm talking about running a research project to find a pandemic capable virus, or as you were saying, make a currently innocuous one nasty, or something like that, and it's like, no, no, no, no, no. I'm assuming that projects like spillover.global are going to create a ranked order list of all the nastiest.
Complete with full genome sequences available in Gen Bank, and that there will be detailed step-by-step protocols to assemble viruses of that family, because for many of them there already are. The question is, can you follow that step-by-step protocol? And of course, get your hands on the synthetic D n A, and then they'll say, oh yeah, well maybe.
I suppose it would take resources because you know, you need access to a lab and if you don't have it probably costs you a good $50,000 in buying that used equipment to get one plus ordering the synthetic D n A yourself, but put it all together and it looks like, yeah, it's pretty clear that many people could do it today.
And what's funny is how people divide on that, because some people say, yeah, okay, but it wouldn't be easy and it would take resources and not many people could do it, which I have to say, okay, well when you crunch the numbers, even if you assigned a pretty low probability that any given virus will actually take off and cause a pandemic.
Covid killed 20 million people. World.
You cannot get that number by putting it into the NUC map simulator of nuclear detonations with any operational nuclear weapon, you can't even get close to it. So what you're saying is if there's an influenza virus that you think even has a small chance of taking off, the expected number of casualties approaches that of a nuclear detonation in a crowded.
and it's okay that anyone has access to this power. That's what I don't really understand. And of course, the real problem is that there aren't any good candidates today. I mean, that's not a problem. That's our saving grace right now. But we will eventually find them. And my expectation is that we're going to describe them in the scientific literature.
We're going to debate them. We're gonna argue would it really. Does the sing really have the capability? Does it not? We're going to pu publish the experiments intended to find out the answer until we make it credible. And I don't, I don't see a way to change that. That's what the norms are. We're too used to fighting nature and nature doesn't fight back that way.
Nature doesn't use what we know against us.
[00:21:18] Kevin Folta: It's a really interesting approach on this entire thing. I mean, is there any evidence of any non-state actors or hostile governments that have even started to approach that idea of poking around with scientists to learn more about it? Maybe start down that path?
[00:21:36] Kevin Esvelt: Well, if you browse the internet, you can always find some people who want to do pretty much everything, including kill lots and lots of people. , and if you want of historical examples of people who actually committed mass murder, I already mentioned the uni bomber, but there's worse than that. There's a, there is a guy by the name of Cchi Endo, who was a graduate trained virologist from Kyoto University, and he joined an apocalyptic cult called Oishin.
and they tried to obtain samples of Ebola for use as a biological weapon against civilians, and they failed. And back then that was pre virus synthesis, so they couldn't make it themselves. Nowadays there's a reverse genetics step-by-step protocol for making Ebola from synthetic D n. and I would be stunned if the typical graduate trained virologist specializing in genetic engineering out of Kyoto University could not successfully follow that protocol today, but they failed back then.
So he just ended up assisting them with using chemical weapons that they managed to synthesize against civilians in the Tokyo subways to commit mass murder. And so he and his elk were executed a few years back for doing. , but it's only been 50 years since the dawn of recombinant d n a, since biotech was a thing at all.
And we've already had one person who clearly had the required training such that if he lived today, he'd be able to make many of the viruses that we're talking about and clearly had the intent to kill as many people as possible. So that's pretty scary. That suggests your baseline risk for training plus intent.
is 2% chance per year because we have n equals at least one in the last 50 years. You could argue about whether he would've succeeded. The typical criticism of biological weapons in the past is, oh, it's all about the delivery. But that's because they were talking about things like anthrax. You have to successfully aerosolize the thing and spray it around.
Now we're talking pandemic agents. You just need to infect people. and a deliberate pandemic would be worse than a natural pandemic or an accidental pandemic involving the exact same virus. If only because anyone competent enough to make the bloody thing in the first place would almost certainly release it across multiple travel hubs, thereby ensuring it spreads everywhere quite a bit faster than the natural or accidental equivalent, and that means less time to develop countermeasures and we might not have enough time to begin.
because the Omicron variant arose in Southern Africa and within a hundred days of sequencing that virus, it had spread to infect a quarter of Americans in nearly half of Europe within a hundred days. How much faster would something like that be if released in multiple airports from the get-go? That's pretty frightening.
We would not have enough time to develop a vaccine for that. Because the stretch goal of the a hundred days mission is to yes, have a vaccine developed, tested, and approved within a hundred days, which is great. We should do that, but we need to do it faster than that because National Biodefense strategy calls for having enough vaccines for every American 130 days after virus sequencing.
So that would be 30 days after Omicron already infected a quarter of Americans. Now imagine that Omicron had been 50%, let. Well, it wouldn't have gotten that far because people wouldn't have gone out. People would be too afraid. But you know, food doesn't get on our tables by itself. Eventually, the water will stop flowing in many places.
We need power or we freeze. . And if you start lacking those things, then people are gonna have to come out and try to find them and take them. And if you don't have law enforcement, then everything falls apart. But if the people supplying all those services aren't willing to go out and risk their lives with N 95 masks offering at best, 95% protection against infection,
then we could lose it. and what scares me is that folks don't seem to have been considering this. That is, we've kind of forgotten what nuclear war means, and even nasty nuclear winter probably isn't the end of human civilization on the southern hemisphere at least even off full on nuclear exchange with the arsenals at the height of a cold War Wouldn't have done that, but a pandemic that actually hit just about everywhere.
we could lose it all. Super unlikely. How many viruses have that kind of lethality and are also highly transmissible, but do you really think we'll never find one given the rate that biotech has been advancing? Or do you really think that we can keep ourselves from describing it? When just a couple months ago, there was a paper in cell.
Published on how a new class of primate, obscure viral family art, arteri viruses are poised for emergence and pandemic potential. So you can get a cell paper by identifying a new virus that you think might have pandemic potential. It's a pretty strong incentive to go out and find it.
[00:27:24] Kevin Folta: Well, this has been a very interesting and sobering first half of the Talking Biotech podcast. We're speaking with Dr. Kevin Selt. He's Associate Professor of Media Arts and Sciences in the M I T Media Lab at M I T, and this is the Talking Biotech Podcast by Col Collabora. And we'll be back with the other half of the story in just a.
And now we're back on the Talking Biotech podcast. We're speaking with Dr. Kevin Esvelt. Kevin is. An associate professor of Media Arts and Sciences at the M I T Media Lab at the Massachusetts Institute of Technology. And you might recognize Kevin from some other Keystone Scientific steps forward.
Maybe the use of CRISPR and gene drives and other goodies. Kevin we started out talking about. The potential threat of being able to order custom d n a and create custom d n a to tweak viruses to potentially use this in a malevolent way. But there are some apparently systems in place to push back against this.
So when you look at mitigation strategies, That are already in place. What's happening right now in major transportation hubs or whatever to surveil for current viral threats?
[00:28:44] Kevin Esvelt: It's a great question. So we know that having a few days warning when a pandemic can spread exponentially can mean a lot. So right now we tend to rely on clinical surveillance that is, you notice people are, if they're sick with something that has strange symptoms or there's a bunch of people with the same symptoms, so you think it might be infectious, then increasingly, cuz the cost of D N A sequencing has gotten so cheap, we throw it on the sequencer and figure out what in the heck it is.
And that's great. That's exactly what we should be doing. But. There is a scenario that still worries me, which is what if you have something that's reasonably fast spreading or at least not slow, maybe respiratory spread or maybe contact but faster than say H I v, but doesn't have those acute phase symptoms.
And instead, like H I V has a pretty long incubation period before it bites you and bites you hard. Would we notice if something like that started spread? Before it got many or most of
[00:29:57] Kevin Folta: us. That's an interesting question. But is all this really boiled down to a need to have maybe a tighter leash on DNA synthesizers or the ability to create nucleic acids that could be potentially used to, like you say, boot up a virus?
And, and spread an infectious agent in a population intentionally
[00:30:22] Kevin Esvelt: well, I think it boils down to, we probably ought to do something about this before we really do become vulnerable, because as previously mentioned, there really aren't a lot of great candidates for causing mayhem with biotech right now.
So, , but we know equally well that we're pretty vulnerable to pandemic agents. We're not so hot at keeping them on a leash. So what we ought to do is delay the day when these things become widely accessible and ensure that we can reliably see something coming, especially if it's that kind of subtle h i v like thing that I mentioned, and ensure that we can shut it down.
And I'm actually quite optimistic about all of these things, so let's just walk through it. DNA n a synthesis is the obvious bottleneck. If you don't have the D n A or some other sample of the virus, you don't have anything to worry about. So obviously we should keep a little bit closer handle on different nasty viruses that we have, but we already do a fair job of that.
I'm sure we can. We can definitely improve it. So it's really the D n A that we need to get a handle on. How can we do that? Well, we need some screening system that will apply to all providers, and if we want 'em to do it and do it reliably, then it probably needs to be free. Because if it costs nothing, then why not adopt it?
Why risk the liability? It's free after all. But if it's gonna be free, it's gotta be fully automated. You can't have any need, any humans in the loop right now. Folks run blast algorithms just similarity, search fuzzy match to try to figure out whether something is hazardous and that's their, you know, final check.
But that creates a f bunch of false alarms that require people with PhDs to look. humans in the loop are expensive and slow. You want to make it free. You gotta fully automate it. So you gotta make the false alarm rate negligible. If you show a, you know, if I were to show you the D n A synthesis order history of, say, a biotech of a biotech firm, that is, then you could probably tell me what their research agenda is.
Do you think given a bunch of DNA sequence orders, ASPO, assuming you're a competitor of that biotech or you're working for pharma or something and you want to know what they've got up their sleeve, if you have a history of all their DNA Synthes orders such that you can just blast or exact match, search each one of them to open databases, you know what genes they've been working with.
You know what organisms, you know what their targets. , you can probably figure out what the research program is, right? So the idea that someone else gets to look at your DNA Synthes orders is pretty darn sensitive. In fact, this is one reason why companies wanna move towards having machines in-house so they don't have to send out their orders to some other party that might have weaknesses in the cyber sense.
So you gotta be privacy, preserving. It's gotta be believable and open source. It's gotta be frictionless so it doesn't delay research. So if you, everyone, so customers are willing to put up with it and it's gotta be comprehensive. It's gotta detect all the known hazards. It's gotta be sensitive. You've gotta, you know, prevent people from ordering a ha you know, a virus within just enough mutations to make it seem like something different while still remaining functional.
And of course you need it to be up to date. People are gonna discover new pandemic class agents. You really want every provider and DNA n a synthesis machine in future to be immediately updated and refuse to make that thing without permission. And lastly, of course, you do need a way for legit scientists who have permission from their relevant biosafety authority to get their hands on the dna n a, they need to do their experiments because as yet, people really haven't killed.
Many people with biotech, in fact, arguably, perhaps no one at all. And it's been a massive net. Good. So we don't want to get in the way of that. And all of these problems are solvable through the magic of better algorithms in cryptography. So that's what we've been working on at a What's now a Swiss foundation called Security, d n a, and it's just software that searches.
Sub sequences of critical hazards and computed functional variants of those sequences so we not only screen for signatures of say, a virus, but also of all the different versions of that particular amino acid or d n a sequence that could plausibly substitute for the original and might otherwise be used to get around.
and it looks like because we are looking for exact matches to these pre-computed functional variants or the exact sequences of the hazard, we can just screen for all the innocent sequences in the database and throw out anything that matches an innocent sequence, so that way you don't really have a false alarm rate.
And. Exact match search is computationally cheap, so you can apply cryptography so that the screen system doing the screening doesn't actually know what the synthesis order is. Comprised of that information never actually leaves the relevant box. , and best of all, it can be integrated into the next generation of D N A synthesis machines.
So we've been working with a number of I G S E firms who are currently testing the alpha version. And our hope is that by making this free in a philanthropically subsidized way, and it's been developed by folks mostly cryptographers from the us, Europe, and China together. So we hope it can be acceptable to just about.
We'll see, of course there's early days. There's another of o other promising initiatives. The Nuclear Threat Initiative has launched a new org IBUs that is trying to lay all the groundwork to get this kind of stuff done and in place. So I'm pretty optimistic that the, this massive security vulnerability of un unscreened DNA synthesis might actually be closed in the next few years.
[00:36:54] Kevin Folta: Now it's a really interesting question end to end because it, it seems like th that this is really not a question of if it's a question of when and no. All the safeguards you can put in place, all the Double checking all of the reporting, all the things that you can do with the DNA n a synthesizer, it seems like somebody will find a way around it and, and really is this, is it really just a question of, I mean, do you personally feel that it's just a question of time before we actually do see a biological threat from an engineered virus?
[00:37:29] Kevin Esvelt: It's pretty likely to happen eventually, but again, the, I think there's a decent chance that by the time it happens, We'll already have defenses in place such that it won't really be able to hurt us because it's not just a matter of, I mean, DNA synthesis screening will delay the day when it happens, right?
It can dramatically reduce the number, certainly of individuals who will be able to do it, because even if you know how to. Assemble DNA N and make a virus. You probably don't know how to break the hardware, lock on the DNA n synthesizer so that you can get the DNA n in the first place. It just makes it sufficiently harder that many, many fewer folks will be able to do it, and especially fewer individuals.
And as soon as, as soon as the malevolent folks have to work with other people, then it gets a lot harder for them. Now you might notice. I haven't said anything about state actors. Obviously nation states can make whatever D N A they want. There's no way we can plausibly stop them, but they also have no incentive to play about with pandemics.
I'm really worried about the lone wolf individuals, and I think we can reduce access to them by quite a lot. We might also be able to do something about delaying. Discovering pandemic capable viruses and publishing them in a ranked order list and shouting about it. But I'm less optimistic there. But if we can put it off, then that means we have time to build a reliable detection system so we see it coming.
So like that subtle threat that I mentioned, it still has to grow exponentially and it still has some unique D N A sequences in it we haven't seen before. . So if you just sequence deeply enough and say, airport wastewater, airplane wastewater especially, and look for exponentially growing sequence fragments, or ones that appear consistently in more and more airports over time, then you know that that's something that's spreading in the pop in the population of folks who are traveling by air and they're gonna get hit either first or second.
So that means we won't be caught off guard. We will know everything biological that is spreading in people. And you could imagine in future we might even be able to extend this to the environment. And once we see it coming, then we can do something about it. And what's the obvious way of solving this problem?
Well, don't breathe contaminated air. Don't make it possible to touch your face when you've contaminated your hands. In other words, you just need to wear a device that feeds you sufficiently clean air. You need really great pandemic proof, p p E, and I don't mean a mask. Mask isn't gonna cut it not on something that's 50% lethal.
You don't wanna trust yourself to a standard style mask. You want a powered air respirator. And if we can get enough of. To all of the essential workers within a few days and ask yourself, can Amazon ship 40 million packages in three days? Of course, this is not a difficult problem. We just need to map all the truly essential workers, figure out how fast we can make the p p e, figure out how much we need a stockpile, of course, decide what's the best design in the first place, and then do it.
But if you're talking. $250 a unit, including distribution costs. Then even if we need 40 million of them, that's 10 billion. It's a pitance in the Pentagon budget. It's even accessible to a generous billionaire if you had to do that, and once you have that, everybody else can shelter in place, not go out, and the essential workers can have perfect P P E and go out there and keep civilization running while we make more P P E.
with everyone doing that transmission will be stopped in countries that can do this at least, and we will be minimally hurt. We can win. But of course, none of us wants to wear P P E all the time, so it'd be even better if we can just stop all transmission of infectious, nasty things between humans, period.
Now this might sound crazy, but it's really the ultimate end goal. . That makes sense because if we think we might be moving into a regime where people can attack us using biology. Well, in cybersecurity, if you allow the adversary to inject information packets into critical systems, you probably have lost.
That's just not a good idea. You just don't let them have access in the first place. . And so with biology, that means we just need to shut down infectious disease. And that seems like it might be hard, but the p p E does it. But without p p e, almost all transmission is indoors and turns out light below 230 nanometers gets absorbed by proteins very, very strongly, meaning it really doesn't penetrate the surface of our skin or eyes, but it's pretty good at frying everything that single cells or viral.
So right now the legal limit for this in the United States is enough to inactivate 90% of aerosolized viruses floating around within a minute. So that's eight times better than an aircraft ventilation system running full blasts. It's pretty. and the preliminary data says you may be able to go 10 times higher or maybe even 50 times higher.
So imagine you're at, I don't know, a rave, which seems like a great way to pass on viruses, close proximity, shouting and so forth. But at least the aerosols would be, 90% of them might be inactivated within a second of leaving the mouth or nose. And of course, it would sterilize surfaces as well. , but it wouldn't harm the people.
So if we could do this and develop LEDs to install it everywhere, it could be the end, not just of pandemic threats, but of the common cold, the flu, R s V, the works, most infectious disease. The hard part for someone like folks like us to accept is that you'll note that this tech, those, neither of those key defenses, don't involve biology at.
And that's cuz bios are just too slow and not reliable enough at the end of the day, against a truly serious threat. The pandemic spreads exponentially and our logistical systems are just not fast enough to keep up. We just can't develop, test, approve manufacturer and distribute a vaccine in time, and we can't make reliable vaccines for every virus.
So there is a solution long term, not just to pandemic threats, but to plausibly all infectious disease. . We just have to get there. But once we do, how awesome will it be speaking as someone whose kids are both sick and I'm sure I'm gonna come down with it tomorrow. .
[00:45:00] Kevin Folta: Well just, you know, I know that there are people who are listening to this because they're attracted to the subject who've never listened to the podcast before, who are screaming.
What about sars Cov two And, and what are your thoughts about its origin?
[00:45:15] Kevin Esvelt: Oh God. Okay. Here's the thing. It is not helpful for us to argue about this. I know everyone desperately wants to know. We don't have enough information to know one way or another. Yeah. It could have spilled over. Yeah, it could have come from a lab.
Yeah, it could have come from a virus hunter. We don't have enough information to know with confidence, and we're probably never going to, because if it's conclusively shown to be spillover from a wet market in Wuhan, China looks. And if it's conclusively shown to result in a lab leak, China looks bad. But if we spend a hard time here arguing over it because there isn't enough information, then we're here arguing and China doesn't look so bad.
Therefore, they're, they have every incentive to avoid giving us sufficient information to know with any confidence. It could have been either. Let's accept that and move on.
[00:46:10] Kevin Folta: Yeah, I, I agree a thousand percent. But on the sociological side, the one thing that we've seen with Covid 19 is that it really has eroded trust in the establishment that monitors public health and.
Provide some sort of public health guidance and where I live right now, they say, oh, it's just a joke. Anyway, it was always a sham. It's, it's over now. You know, don't, there's no, and I wear a mask. I mean, I got, I got a couple situations that require me to keep, keep free of viruses. And so when you see rejection of fundamental issues like germ theory promoted by online influencers, by a network of 65% of false information coming from 12 accounts on Twitter with respect to covid politicians saying it's nothing.
Does this kind of denial really enhance the threat from a legitimate concern that comes? So, in other words, some, some actor generates a a new viral threat. Some, you know, North Korea, whatever we get, we get a flag that this thing is in the airports and we have three sick people stay home while we put on perfect respirators and go out and solve the problem.
What is the likelihood that anyone will abide to public health guidance?
[00:47:30] Kevin Esvelt: Well, they won't initially, but if it's highly lethal and obviously so that will change pretty rapidly and so a bunch of people will die needlessly new to early due to early spread. But I think people will come around if it's obviously highly lethal.
What really concerns me is that second scenario I mentioned where it has a long incubation period before anything nasty happens. Cuz there we might be in a situation where, Build our detection system based on exponential growth or some other pathogen agnostic mechanism. We see the thing coming. We recognize that it's a threat.
Maybe even we can show that it's engineered, which might be one way past the politi politicization, if we can show that it's engineered. But the, that's the hard one. How do you convince people to take precautions when nobody's in the hospital and nobody's died yet? That's why I want that 2 22 nanometer indoor germicidal light because it's a passive defense.
People don't have to believe in the threat and it'll still fry it out of the air before it infects 'em.
[00:48:37] Kevin Folta: With all of this in mind, has any state actor really started to generate, or any evidence that they've created, pandemic level engineered virus?
[00:48:48] Kevin Esvelt: Well, we definitely have confirmation that the Soviet Union was experimenting with smallpox and that they may well have selected or engineered it somehow for greater lethality.
And we know this because there was a, there was a leak. During a test on the island of a Rosk A, which is where they were testing their biological weapons, a boat sailed nearer to the island than it was supposed to. It was supposed to, you know, it was mirrored 10 or so kilometers away, but they were interested in, you know, explosive based delivery of biological weapons pandemic class, small.
and some people on the ship became infected and they shut down the trains. They isolated everyone and they vaccinated everyone. But a number of people were infected, and of the folks who were not vaccinated, all of them died. But smallpox is normally only 30%. Let. So it could have been a cystical fluke or they might have made a nastier version.
We don't actually know.
[00:50:07] Kevin Folta: I understand where you're coming from. You've been studying this, you've watched this carefully, and it, it is a horribly interesting topic and you know, what are your biases you bring to the table and how could you just be completely wrong about this?
[00:50:21] Kevin Esvelt: Well, the main way I can be wrong is it's always a low probability thing, right?
Even if we find a bunch of really nasty new pandemics and we advertise that they exist because we're trying to stop 'em from being from spilling over and causing natural pandemics, and we make the ranked order list, maybe no one would actually do it. State actors don't have any particular interest in doing.
maybe Cido was a fluke. Maybe it's harder than it seems to us in the lab who at least can do it for the easier viruses maybe. But I think probably the biggest disclosure is that I'm biased because as you mentioned, I played a major role in inventing CRISPR based gene. And that was really the, perhaps the first biotechnology that really will spread on its own without human intervention that we're pretty confident that it will in fact, do this.
No one's released yet in the wild, so we're not sure, but there are quite a few examples in cage populations where it works and including crashing. For one's designed to crash populations of malaria mosquitoes, for example. And so I kinda hold myself morally responsible for that technology, and I first got into this by calling for, in that case, transparency, community guidance and universal use of laboratory safeguards to prevent accidents because we know that it took a single death in a clinical trial, the setback.
Gene therapy by 10 years. And gene drive could be a very useful tool against horrific diseases like malaria and schistosomiasis. But if there's an accident and the headlines say scientists accidentally turn entire species into GMOs, is CRISPR to blame? . Well, that won't be any good for biotech, for trust in science, and it certainly won't be good for the prospects of actually using Gene Drive to eradicate these diseases, which means that if there was something that I could do as an inventor to reduce the risk of an accident and I didn't do it, then I would be morally responsible for the consequences, which would involve quite a lot of children dead of malaria.
in fact, a 1% chance of a decade long delay equates to on the order of 25,000 dead kits. And that does things to your sense of morality when you start computing things that way. So I spent quite a, a lot of effort advocating for safeguards and transparency and trying to ensure that the technology could be accepted by going out and talking to communities and.
Starting new projects that didn't involve Gene Drive because the initial community didn't want it and we picked islands so they wouldn't necessarily need it. And trying to figure out how to do it in a way that makes it most likely that the technology will be both safe and accepted.
So on the one hand, I'm the person who can speak with some authority on exponential bioTE. On the other hand, I'm probably the single most biased individual in the world when it comes to the risks stemming from exponential bio technologies. So now that I'm pointing out, hey, there's going to be others, gene Drive was really just the beginning and it was a fairly safe beginning because it's slow, obvious, and easily blocked.
Pandemics are not necessarily, you should be at least a little suspicious that I am biased because I had to deal with that.
[00:54:22] Kevin Folta: All of this has been really very intriguing and I think it, it's a, a sobering flip side of what is normally a pretty sunny podcast. We, you know, we talk about all the great innovations that occur, and it's always good to talk about how these things might go wrong, and I really appreciate your time on this today.
If people wanted to learn more about your program and what you do and what you. In the other facets of your research program, where would
[00:54:49] Kevin Esvelt: they look? So our group is called Sculpting Evolution, and if you Google that, you will find us.
[00:54:56] Kevin Folta: That's sculpting Evolution. Are you on Twitter or anything like that?
[00:55:00] Kevin Esvelt: Yeah, I'm at at KSL at Twitter and we have our webpage at m i t and. Not at m i t and elsewhere, but also I also happen to be Google unique, so it's not hard to find things. I've done Google
[00:55:14] Kevin Folta: Unique. I haven't heard
[00:55:15] Kevin Esvelt: that one before. I Oh, just means your, yeah, just your first and last name are unique. Yeah, I
[00:55:19] Kevin Folta: know.
I'm Google dead. That's from false information. Yeah, but that, that, you know, so that, but that's, that's why you keep doing the good stuff, right? You're trying to , always trying to outrun a sawmill, blade .
[00:55:33] Kevin Esvelt: It's always coming by you, but that's why. Honestly don't post very much on Twitter.
[00:55:38] Kevin Folta: Yeah, no. Especially in your business.
I could see why not. But Dr. Kevin Feld, thank you very much for your time on this today. I really appreciate it. And as time rolls on, I hope you can come back again and talk to us again about this really important area that we need to be a aware of. And I really appreciate your time on this. Thank you.
Thanks, Kevin. Cheers. And as always, thank you very much for listening to another episode of Talking Biotech podcast. Share this one with a friend, you know, or someone in the family who you think should understand that there are important set of risks and considerations for whenever we forward. A good technology that there's potential ways that it can go wrong, and it's important for us to understand what those are, to brainstorm on them and think about them before somebody else does, so that we're ready for when the threats come along.
This is a Talking Biotech podcast, and we'll talk to you again next week.