We’re developing a blueprint for evaluating the risk that a large language model (LLM) could aid someone in creating a biological threat. In an evaluation involving both biology experts and students, we found that GPT-4 provides at most a mild uplift in biological threat creation accuracy. While this uplift is not large enough to be conclusive, our finding is a starting point for continued research and community deliberation.
  • skillissuer@discuss.tchncs.deEnglish
    4·
    10 months ago

    The equipment and ppe for bio weapons and chemical weapons of the same health hazard is about the same.

    well i think it strongly depends on your threat model. consider small leak of sarin: effects are detectable in seconds to minutes, antidote is readily available, cheap and specific, and sarin poisoning is not transmissible. in case of, say, plague, you won’t know what’s going on for days in which time there’s already a risk of infecting some random passerby, which is highly suboptimal if you want to stay covert

    In either case your logic is relying on a threatening actor to not have any education

    it’s not that i assume no education, it’s that i (as an organic chemist) wouldn’t trust crystallographer or electrochemist with synthesis of something like sarin. even with all required PPE and other precautions, your recruiting poll drops from about 100% (IED carrying child soldier) to maybe 1-0.1%? and that’s even before you consider that some of these highly specialized chemical weapons people are already on military payroll, or are surveilled precisely for this reason

    you’re underestimating cost of this entire enterprise, which even at lab scale could easily go into hundreds of thousands to million dollar range. you’re underestimating how hard it is even when you have everything provided - look at iraqi chemical weapons program. with no need to stay particularly covert they were only able to manufacture mustard gas of useful quality that could be stored; their mid tier chemical weapon sarin was at something like 30% purity and had very short shelf life; their vx was so dirty it was straight up useless

    for some weird reason you’re assuming that whatever chemistry you want to do, it works on the first try. it won’t; it never does, and even if it did, you have to make sure you’ve got the right stuff. this makes synthesis only half of the problem, because there’s still purification and analysis

    you seem to ignore that even in the paper that you cite, anyone that doesn’t have to do chemistry, doesn’t. (by that i mean performing some reaction, that generates side products, and so requires purification, analysis, and generates waste stream). doing chemistry means generating waste and need of its safe-ish disposal; it means getting considerable PPE; it means getting precursors, maybe in large amounts; all of that might move you from government watch list to government act list.

    talib doesn’t do chemistry when he makes an IED, because melting down contents of TM-62s or UXO found in nearby field isn’t chemistry; unabomber stuffing match heads in a pipe isn’t chemistry; stealing cylinders of chlorine (bulk of fatalities in that paper) and putting it in a car bomb isn’t chemistry. chlorine is not something you make and put in cylinders, because it’s relatively hard, uses large amounts of energy, leaves considerable waste/side product stream, and you can order it on aliexpress. same goes for sulfur mustard, i’m pretty sure most of incidents happened in syrian civil war and ultimately this stuff can be traced to syrian or iraqi chemical weapons program

    most of these problems, but especially making sure you’ve got the right stuff, are much harder for living organisms than for clearly identifiable, publicly known compounds. and we’re still nowhere close to the point where llm gets potentially useful. no, getting B in high school biology and relying on gpt4 and scihub to get all the way up there doesn’t count. chatgpt writing out rna sequence to be printed out, engineered into a bacterium and spread by a cultist, all done by mail order and or in garage is scenario completely detached from any pretense of being realistic

    beakers, Bunsen burners, and filters

    this tells me that you’ve ended all contact with chemistry on (classical, aqueous) qualitative inorganic analysis, because if you tried to cook anything on bunsen burner in organic lab, that’d be pretty hard considering there are none in flammables area. have you considered that you’re severely out of your depth and got caught in openai’s fear based hype-marketing?

    e: if you want to isolate anthrax from dirt, you’ll have many more problems than that, especially with “getting the right stuff” part. there are places where anthrax is endemic, but if step 1 involves catching diseased marmot in southern mongolia or deer in eastern siberia, this devolves straight into rube goldberg machine of mass destruction area

    • skillissuer@discuss.tchncs.deEnglish
      2·
      10 months ago

      in context of Iraqi insurgency even things like EFP plates were industrially manufactured in Iran and shipped there by their special forces, even that it’s just a chunk of copper plate pressed in shape of shallow cone. same in Afghanistan, where friendly CIA/ISI agent, or friendly black market weapons trader depending on period would provide them with explosives, fuzes, communication hardware, training and some modern weapons up to and including FIM-92 Stingers

    • Umbrias@beehaw.orgEnglish
      1·
      10 months ago

      Re: ppe

      Sure, the outcomes are different, but the scale is too. The scale of a chemical weapons program is necessarily higher from a hazard point of view due to the sheer volume of material. The specifics make that messy though, yes, any particular pathogen would want differing levels of ppe.

      Re: precursors

      Right, the part of the point I’ve made about bio weapons is that spotting the precursors is very difficult, because a normal bio lab needs roughly the same stuff a weapons bio lab does.

      I don’t disagree that many of the chemical weapons used in Syria may be from larger chemical weapons programs. But that doesn’t mean lab scale ones don’t also exist.

      Re: ended contact with chemistry

      Not everybody is trying to posture. The point wasn’t to show off a magnificent knowledge of lab equipment, but to demonstrate the similarity at a high level.

      Re: llms for biology

      Ehhhh there are plenty of research applications of llms at the moment and at least one is in use for generating candidate synthetic compounds to test. It’s not exactly the most painful thing to setup either, but no if you were to try to make a bio weapon today with llm tools (and other machine learning tools) alone it would go poorly. Ten, twenty years from now I’m not so sure, the prediction horizon is tiny.

      Re: caught in openai fear

      Why would I consider that when my opinions on bio weapons and CBRN are wholly unrelated to openais garbage? I didn’t even know openai cared about CBRN before today and I fully expect it’s just cash grabbing.

      People can abuse and misinterpret real concepts in a way that makes them seem absurd.

      Yes in practice anthrax is nontrivial. But folks here also seem to think any of this is magically impossible, and not something that dedicated people can reasonably do with fewer resources by the day. Which by the way is great, the surge of amateur micro bio is great, we’re learning a lot, we’re getting very smart hobbyists contributing in unexpected ways.

      • skillissuer@discuss.tchncs.deEnglish
        3·
        10 months ago

        right, sure. there are very few labs that require select agents, and running an anthrax vaccine program out of your backyard is a hobby i haven’t heard of yet

        lab scale cws are just that, lab scale. multi-kg amounts are not lab scale, and unless you’re running around with suspiciously modified umbrella, sub-gram amounts aren’t casualty-producing

        The scale of a chemical weapons program is necessarily higher from a hazard point of view due to the sheer volume of material

        you can’t grow some random bacteria in high concentration, you’re looking at tens to thousands of liters of fermenter volume just to get anything useful (then you need to purify it, dispose of now pretty hazardous waste, and lyophilize all the output, it gets expensive too)

        The point wasn’t to show off a magnificent knowledge of lab equipment, but to demonstrate the similarity at a high level.

        for the reasons i’ve pointed out before, there’s none to very little similarity

        Ehhhh there are plenty of research applications of llms at the moment and at least one is in use for generating candidate synthetic compounds to test. It’s not exactly the most painful thing to setup either, but no if you were to try to make a bio weapon today with llm tools (and other machine learning tools) alone it would go poorly.

        it’s nice that you mention it, because i’ve witnessed some “ai-driven” drug development firsthand during early covid. despite having access to xrd data from fragment screening and antiviral activity measurements and making custom ai just for this one protein, the actual lead that survived development to clinical stage was completely and entirely made by human medchemists, atom by atom, and didn’t even include one pocket that was important in binding of that compound (but involving that pocket was a good idea in principle, because there are potent compounds that do that), and that despite these ai-generated compounds amounted something like 2/3 of all tested for potency. but you won’t find any of that on that startup’s page anymore, oh no, this scares away vcs.

        Ten, twenty years from now I’m not so sure, the prediction horizon is tiny.

        i’m equally sure that it’ll go poorly then too, because this is not a problem you can simulate your way out of and some real world data would need to get input there, and that data is restricted

        But folks here also seem to think any of this is magically impossible, and not something that dedicated people can reasonably do with fewer resources by the day

        yeah nah again. lately (june 2023) there was some fucker in norway that got caught making ricin (which i would argue is more of chemical weapon), because he got poisoned in the process, with zero fatalities. [1] around the same time single terrorist incident generated about the same number of casualties and much more fatalities than all of these “bw terrorism” incidents combined. [2] this doesn’t make me think that bw are a credible threat, at least compared to usual conventional weapons, outside of nation state level actors

        at no point you have answered the problem of analysis. this is what generates most of costs in lab, and i see no way how llm can tell you how pure a compound is, what is it, or what kind of bacteria you’ve just grown and whether it’s lethal and how transmissible. if you have known-lethal sample (load-bearing assumption) you can grow just this and at no point gpt4 will help you, and if you don’t, you need to test it anyway, and good luck doing that covertly if you’re not a state level actor. you also run into the same problem with cws, but at least you can compare some spectra with known literature ones. at no point you have shown how llms can expedite any of this

        you don’t come here with convincing arguments, you don’t have any reasonable data backing your arguments, and i hope you’ll find something more productive to do over the rest of the weekend. i remain unconvinced that bw and even cw terrorism is anything else than movie plot idea and its promise is a massive bait against particular sector of extremists