[00:01.680 --> 00:03.020]  So my name is Ed.
[00:03.320 --> 00:04.480]  Thank you all so much for coming.
[00:05.320 --> 00:14.380]  Under normal circumstances, I would be up on my feet and kind of giving a bit more of a show, but I was doing trigonometry with my foot last night and broke a Toa.
[00:14.860 --> 00:20.300]  So I'm trying to stay off of it and I'll be sitting down for as long as I can manage it.
[00:20.360 --> 00:23.900]  If I get up and start dancing or whatever, it's normal.
[00:24.220 --> 00:25.540]  You don't need to call anybody.
[00:26.680 --> 00:42.060]  So this is a talk about export controls and what happens at the border between our country, the United States and other countries when you are trying to transport technologies for commerce or fun.
[00:43.460 --> 00:45.360]  A little bit about myself.
[00:45.380 --> 00:46.540]  I'm an attorney.
[00:46.980 --> 00:49.080]  I'm an attorney in the state of New York.
[00:49.080 --> 00:50.560]  I am a patent attorney.
[00:52.680 --> 00:55.460]  This is not my usual field of law.
[00:56.040 --> 01:00.460]  I've kind of gotten really interested in it in the past, let's say, year.
[01:00.920 --> 01:08.900]  And I'll get into how this dovetails with my usual practice later on, but just know that I have a background in physics.
[01:08.900 --> 01:11.280]  I work with technology a lot.
[01:11.500 --> 01:18.160]  These days, I do a lot of AI, just because AI has swallowed all invention.
[01:18.640 --> 01:33.360]  Everything that comes across my desk has some kind of AI angle now and it's getting a little samey, which is why I've been interested in quantum computing and other things, because you've got to, you know, vary things a little bit to stay sane.
[01:35.060 --> 01:37.900]  And as you can see from the pictures, I'm a very serious person.
[01:38.480 --> 01:43.360]  That picture specifically, you don't have to look at the other pictures, but they'll give you a bit of flavor.
[01:45.360 --> 01:53.840]  Oh, and I promised I would give a shout out to my children, Abby, Edward, and August, who are watching, hopefully via the live stream.
[01:54.340 --> 01:58.860]  And they will let me know how I did later on.
[01:59.920 --> 02:02.200]  So, let's talk about export controls.
[02:02.620 --> 02:03.640]  What is an export control?
[02:03.760 --> 02:06.980]  This is super dry stuff, unless you're a very specific kind of nerd.
[02:08.020 --> 02:09.400]  I think it's very interesting.
[02:09.560 --> 02:11.440]  You might not, but we'll try and liven it up.
[02:11.620 --> 02:21.840]  And export control is a law or a regulation that restricts taking technologies or information out of the country.
[02:22.680 --> 02:27.060]  And it covers, you know, what you normally think about as exports, right?
[02:27.240 --> 02:30.120]  Physical goods, it definitely covers.
[02:30.120 --> 02:32.620]  But it also covers sort of inchoate things.
[02:32.980 --> 02:35.820]  So, it covers services, for example.
[02:36.020 --> 02:38.620]  It can cover software in some circumstances.
[02:38.940 --> 02:43.880]  The definition of what a technology is, I've copied on the right here.
[02:44.000 --> 02:51.940]  It's information necessary for the development, production, use, operation, installation, maintenance, repair, overhaul, or refurbishing of an item.
[02:51.940 --> 02:55.340]  So, it's a very broad definition of what a technology is.
[02:55.500 --> 02:58.260]  It's information that lets you do a thing.
[02:58.260 --> 03:15.220]  And the export control rules, laws, regulations, they define what kinds of things in very, very dry detail exactly you are not allowed to take out of the country.
[03:16.140 --> 03:20.260]  So, I'd like for a little bit of interactivity if we can do it.
[03:20.980 --> 03:27.160]  One of the things I want to talk about here is why we have export controls, and the policy reasons behind them.
[03:27.320 --> 03:34.880]  Because they're, like all matters of governance, they're kind of squishy, right?
[03:35.040 --> 03:40.740]  We have these rules for a reason, but really we have multiple reasons, and sometimes they conflict with each other.
[03:40.740 --> 03:47.880]  So, if I could get somebody to shout out, what is a reason we might not want someone to export technology from the country?
[03:48.960 --> 03:51.620]  I'm going to make it re...
[03:52.440 --> 03:53.680]  National security!
[03:54.400 --> 03:59.920]  We want to prevent the proliferation of certain kinds of technology.
[04:00.440 --> 04:01.960]  Weapons, right?
[04:01.960 --> 04:04.880]  We don't want everybody to have nuclear weapons.
[04:05.480 --> 04:11.420]  And so, nuclear energy is a big part of what we control the export of.
[04:11.620 --> 04:12.200]  But it...
[04:12.740 --> 04:17.540]  There are a lot of things that really get covered under this idea of national security.
[04:17.760 --> 04:20.340]  And they're not all directly weapon related.
[04:20.600 --> 04:23.020]  So, we'll get to some of the weird ones later on.
[04:23.020 --> 04:27.060]  But another option is economic priorities, right?
[04:27.940 --> 04:35.460]  The United States is out there competing with every other country on the planet when it comes to technology.
[04:35.460 --> 04:37.600]  We like to think we're doing a pretty good job.
[04:37.740 --> 04:44.780]  But part of that is we want to maintain our dominance in certain fields by preventing those technologies from weakening.
[04:44.940 --> 04:52.920]  Now, several years ago, many years ago, I gave a talk here, I guess at the Hotel Pennsylvania, about trade secret law.
[04:52.920 --> 04:56.380]  Which has a lot of this similar elements to it.
[04:58.540 --> 05:02.780]  It's really hard to keep technologies in a bottle, right?
[05:02.920 --> 05:04.040]  And once they're out, they're out.
[05:04.160 --> 05:05.600]  And you can't do anything about it.
[05:05.860 --> 05:13.200]  And so, this idea of maintaining these economic priorities, it's really just about slowing down the development of other people.
[05:13.340 --> 05:19.300]  And sometimes, maintaining a dependency on the United States, right?
[05:19.300 --> 05:26.740]  If it's cheaper for a country to buy our stuff than to develop themselves, we have to balance that against...
[05:26.740 --> 05:29.300]  Well, we're kind of giving them the ability...
[05:30.980 --> 05:31.300]  Right?
[05:31.880 --> 05:33.060]  There's a sweet spot there.
[05:33.140 --> 05:35.580]  And it takes a lot of time to figure out exactly what that is.
[05:36.220 --> 05:39.100]  And then there are other things like foreign policy, right?
[05:39.240 --> 05:39.660]  Sanctions.
[05:39.660 --> 05:41.600]  I've got the North Korea flag here.
[05:41.720 --> 05:43.220]  I've got the Iranian flag here.
[05:43.360 --> 05:45.120]  They're both under heavy sanctions.
[05:45.440 --> 05:52.100]  That limit what the United States is willing to give them because they're bad actors.
[05:52.680 --> 05:52.960]  Right?
[05:53.220 --> 05:54.780]  Part of this is human rights.
[05:55.000 --> 06:04.940]  That's been a relatively recent development that human rights have been an explicit part of our policy behind export controls.
[06:04.940 --> 06:11.360]  Where we say, you know what, this device, whatever it is, this technology, has the potential for abuse.
[06:12.000 --> 06:16.980]  And so we are going to limit export of this technology.
[06:16.980 --> 06:24.400]  Not for national security reasons, but because we feel bad about the people in these other countries that this is going to be used against.
[06:24.800 --> 06:28.220]  Which, and we'll get into it later, gets super hypocritical.
[06:29.700 --> 06:33.300]  Alright, so what are the penalties for violating export controls?
[06:33.300 --> 06:34.840]  That's right, straight to jail.
[06:35.600 --> 06:38.920]  You can be... and there are two different levels.
[06:39.080 --> 06:42.680]  There's administrative, which is like civil level, and there's criminal.
[06:43.020 --> 06:47.640]  Criminal means up to 20 years in prison, up to a million dollars per violation.
[06:48.720 --> 06:58.660]  Administrative is a little more mild, in that it's only, you know, 300,000 or twice the value of whatever you made selling it to whomever.
[06:59.340 --> 07:01.620]  And denial of export privileges.
[07:02.360 --> 07:02.440]  Okay?
[07:02.720 --> 07:04.280]  So these penalties get pretty severe.
[07:04.760 --> 07:10.360]  The difference between criminal and civil in this world is a matter of willfulness.
[07:10.780 --> 07:14.640]  For the lawyers in the room, willful means you really meant to do it.
[07:14.700 --> 07:15.620]  You knew what you were doing.
[07:15.840 --> 07:19.360]  You knew what you were doing was wrong and you did it anyway.
[07:19.620 --> 07:20.000]  Right?
[07:21.680 --> 07:25.120]  When it's not willful, it's just like, yeah, I sold him that.
[07:25.120 --> 07:26.480]  I didn't know it was bad.
[07:26.840 --> 07:27.780]  What's the big deal?
[07:28.080 --> 07:30.140]  Then you get dinged, but you don't go to jail.
[07:32.600 --> 07:34.560]  So let's get into some examples.
[07:34.860 --> 07:38.060]  We're going to start with a historical perspective of export control.
[07:38.200 --> 07:42.660]  These are, you know, based in things that I saw going on when I was coming up.
[07:43.280 --> 07:45.360]  A big one in the 90s was encryption.
[07:45.620 --> 07:52.220]  And it's still a big one, but it was really potent in the 90s because encryption was kind of regarded as munitions.
[07:52.220 --> 07:52.880]  Right?
[07:53.000 --> 07:56.980]  This was very squarely in the national security bucket.
[08:00.150 --> 08:06.190]  And so the 90s were also when encryption started to really gain some traction.
[08:06.610 --> 08:09.730]  We had PGP, which was released in 1991.
[08:10.750 --> 08:14.510]  SSL around 95 or 96, which was developed by Netscape.
[08:14.930 --> 08:20.430]  These became, SSL in particular, became a foundational technology for how we use the Internet.
[08:20.430 --> 08:25.850]  Because now it became possible to conduct commerce in a relatively secure fashion.
[08:27.390 --> 08:37.670]  But, and the picture I have up here is actually of a Netscape installed diskette that is marked U.S. and Canada only, not for export.
[08:37.950 --> 08:49.030]  And that is specifically because the export restrictions put a cap on how strong your encryption could be if it was leaving the country.
[08:49.030 --> 08:58.470]  Now, because it was, you know, relatively unrestricted, most people just used the international version of Netscape, which had very weak cryptography.
[08:58.730 --> 09:06.570]  There was a big push around this time to start loosening up the restrictions on encryption.
[09:06.950 --> 09:10.070]  Because it was limiting commerce domestically, right?
[09:10.170 --> 09:14.750]  People couldn't trust that they were communicating securely online with their customers.
[09:15.870 --> 09:19.790]  There were also a couple of interesting lawsuits that happened at the same time.
[09:20.190 --> 09:26.770]  There were, they were making the case that source code is expression.
[09:27.330 --> 09:29.670]  That it is protected free speech.
[09:30.170 --> 09:39.190]  And those cases actually prevailed against the relatively strict encryption export rules at the time.
[09:40.290 --> 09:43.950]  But, you know, I don't know if they would necessarily hold up today.
[09:44.350 --> 09:51.310]  Because I think that they gave a very generous interpretation of what source code is in terms of expression.
[09:51.630 --> 09:55.090]  I think that it's not to get too deep in the weeds.
[09:55.250 --> 09:56.950]  It never went to the Supreme Court.
[09:57.230 --> 10:02.270]  These were decided by courts of appeals on the federal level, but never went all the way up.
[10:03.010 --> 10:07.750]  And so as a result, who knows how that would play out if it got challenged today.
[10:07.910 --> 10:12.610]  But one way or another, by the end of the 90s, export controls had loosened up.
[10:13.350 --> 10:21.750]  Encryption was available for, you know, to use for Amazon and for all the other web marketplaces that started sprouting up.
[10:22.730 --> 10:29.850]  Along this timeline though, there was kind of a, there was a strong feeling of this is nonsense, right?
[10:29.850 --> 10:39.270]  Because this technology was so easy to capture and so easy to export intentionally or unintentionally.
[10:39.430 --> 10:40.850]  There were t-shirts made.
[10:40.990 --> 10:48.470]  I've got the, the source code here of a Pearl implementation of RSA that violated the export controls, right?
[10:49.150 --> 10:55.170]  Anybody wearing that t-shirt who left the country or just carried the t-shirt in their, you know, in their, in their luggage.
[10:55.170 --> 10:59.410]  They were violating export controls by taking this technology outside of the country.
[10:59.990 --> 11:01.730]  Obviously that's loosened up a lot.
[11:01.890 --> 11:11.510]  But encryption remains one of the more limited forms of software because it has this potential for dual use, right?
[11:11.710 --> 11:13.910]  And this is a concept that we're going to run into a bunch.
[11:14.010 --> 11:19.410]  The idea that something can be used for benign purposes, but also for shady purposes.
[11:20.550 --> 11:21.630]  Here's another example.
[11:21.810 --> 11:24.150]  And, and, and I should, I should be clear right now.
[11:24.310 --> 11:26.490]  This is not just something that the United States does.
[11:26.630 --> 11:29.250]  This is a Japanese example of the PlayStation 2.
[11:29.470 --> 11:35.190]  When the PlayStation 2 was being, you know, the, the, the press was coming out for the PlayStation 2.
[11:35.450 --> 11:41.030]  The government of Japan investigated whether they needed to restrict the export of this.
[11:41.030 --> 11:50.550]  The PlayStation 2 had a graphics processing chip and the Japanese government was concerned that that chip could be used for missile guidance.
[11:51.090 --> 11:54.210]  That graphics processing is a thing that guided missiles do.
[11:54.350 --> 12:01.990]  And well, if you're just selling these for super cheap relative to military hardware, people could buy them, build their own, you know, guided missiles.
[12:01.990 --> 12:05.870]  In the end, they did allow the export of the PlayStation 2.
[12:06.450 --> 12:13.270]  Um, with the exception of certain countries, even though eventually Iraq did get a whole bunch of them on their hands.
[12:13.410 --> 12:15.710]  I'm not even sure they did anything with them.
[12:16.130 --> 12:21.130]  They might've just bought into the hype and be like, oh man, I need to get some of these missile games.
[12:21.330 --> 12:23.170]  And they got them like, I don't know.
[12:24.070 --> 12:25.350]  What do we do with these now?
[12:25.510 --> 12:29.330]  But as far as anybody can tell, nothing ever happened from Iraq buying them.
[12:29.330 --> 12:34.430]  Um, life jackets, export controls on life jackets.
[12:34.670 --> 12:40.370]  This is another dual use item because while yes, they're great for marine safety for civilians.
[12:40.550 --> 12:46.230]  They're also great for marine safety for, you know, people on military boats, right?
[12:46.370 --> 12:53.290]  If you start digging into the export control rules, you are going to find all kinds of wacky stuff.
[12:55.210 --> 12:56.970]  And this is just to be clear.
[12:57.190 --> 12:57.430]  Yes.
[12:57.590 --> 13:02.610]  I asked AI to make a picture of a mean guy wearing a life vest.
[13:02.630 --> 13:04.450]  I want to disclaim that up front.
[13:04.630 --> 13:06.510]  I do not own the copyright to that image.
[13:08.110 --> 13:17.390]  Um, and then one other one I wanted to flag for your attention before we, you know, move on is voice print identification and analysis equipment.
[13:17.390 --> 13:23.850]  This is flagged specifically in the rules as being there for human rights reasons.
[13:24.710 --> 13:25.350]  Okay.
[13:25.570 --> 13:35.850]  This is something that our government decided we don't want to sell to questionable places because they might use it to be mean to their own citizens.
[13:35.850 --> 13:55.930]  And on the, on the side here, I've got this bit from the rules that says, um, they'll generally be okay with you exporting voice print identification and analysis equipment, unless there is civil disorder in the country, or there is a risk that items will be used to violate or abuse human rights.
[13:56.150 --> 14:02.830]  And I think it's interesting that we take this strong position on, on technologies like this.
[14:02.830 --> 14:06.570]  There's similar tech, uh, restriction on fingerprinting technology, for example.
[14:07.490 --> 14:11.990]  And we have no concerns about the potential abuses within our own country.
[14:12.150 --> 14:18.350]  We're very blase about how, you know, how good, how we must be the good guys when it comes to this kind of a thing.
[14:20.310 --> 14:25.830]  So let's talk about what is not included when we talk about technology, right?
[14:25.950 --> 14:28.510]  Because technology is a very broad term, right?
[14:28.750 --> 14:31.790]  Covers basically any kind of instruction and how to do a thing.
[14:31.790 --> 14:38.050]  But, publicly available information, public act, uh, published academic papers, for example.
[14:38.690 --> 14:40.870]  You know, manuals, things of that nature.
[14:41.490 --> 14:42.710]  They're, they're available.
[14:42.990 --> 14:48.130]  You could go to a library, you know, you could get on the Internet and download all kinds of information.
[14:48.510 --> 14:51.350]  That's not restricted because it's already out there.
[14:52.250 --> 14:55.030]  Fundamental research is another big piece of this.
[14:55.450 --> 15:00.530]  Now, earlier I said there are people who spend their whole careers thinking about what the right balance is.
[15:00.530 --> 15:01.110]  Right?
[15:01.210 --> 15:03.070]  There's a balance of priorities here.
[15:03.470 --> 15:13.030]  The reason for the exclusion for fundamental research is because there needs to be a free flow of information to advance the state of the art, right?
[15:13.030 --> 15:21.010]  The people in the United States, yeah, we think we've got great smart people, but there are great smart people in other places who are doing great work.
[15:21.150 --> 15:28.170]  And they need to be talking to each other or else we are going to very quickly get left behind by everybody else who's talking to everybody else.
[15:28.910 --> 15:31.510]  Um, so fundamental research, what does that mean?
[15:31.510 --> 15:35.750]  That means, generally speaking, something that you're not claiming a proprietary right in.
[15:36.490 --> 15:36.970]  Right?
[15:37.170 --> 15:48.590]  So if you're in a university and you have publicly funded information, you're going to publish all your results, you're conducting fundamental research, you don't have to worry about the export controls on what you're doing.
[15:49.510 --> 15:55.390]  If now you're working for a private company and that private company says, well, we're going to file a patent on that.
[15:55.590 --> 15:57.710]  So don't talk about it.
[15:58.450 --> 16:06.450]  Now that information, that research you're doing is not fundamental research, and it's not publicly available information.
[16:06.610 --> 16:10.830]  That work counts and has to be considered under these rules.
[16:11.830 --> 16:28.590]  And then finally, most software, and like I alluded to earlier, there's a tension between the first amendment protection that software code is granted, and the fact that software does stuff, and sometimes software can do bad stuff, and we need to have some kind of grip on that.
[16:28.990 --> 16:48.010]  And so there is, uh, encryption is really the prime example of that, where we've established very clearly in the courts that encryption has first amendment protections, but you still do need to have a license from the government if you are going to export certain software.
[16:53.930 --> 17:01.110]  That license, by the way, is really where the legal wheels are turning.
[17:01.430 --> 17:01.750]  Right?
[17:01.890 --> 17:18.530]  So if you are a company or an inventor or whatever you are, and you want to sell something abroad, and what you want to sell is on this list of restricted items, and you want to sell it to one of the countries you're not supposed to sell it to, you need to get a license from the government to do so.
[17:18.770 --> 17:32.010]  Where somebody in the government, uh, at the Bureau of Industry and something that starts with an S, is going to review your application and say yay or nay based on whatever the factors are of your particular needs.
[17:33.050 --> 17:33.130]  Okay?
[17:34.370 --> 17:39.690]  One of the things that factors into this is that it's actually really easy to export stuff.
[17:39.690 --> 17:45.770]  You don't have to send a physical product overseas for it to count as an export.
[17:45.950 --> 17:49.030]  And that's where we enter the world of deemed exports.
[17:49.450 --> 17:52.470]  So imagine you have a colleague who is from China.
[17:53.030 --> 17:53.570]  Okay?
[17:54.350 --> 18:00.090]  Chinese citizenship, been here for years, doing great work in the field of whatever.
[18:00.810 --> 18:01.050]  Okay?
[18:01.770 --> 18:05.450]  Your field is, you know, you don't work on the same stuff though.
[18:05.690 --> 18:08.230]  You work on restricted things.
[18:08.710 --> 18:09.710]  They don't.
[18:10.850 --> 18:18.510]  They are not allowed to learn about your restricted technology, because it would be deemed an export back to China.
[18:18.510 --> 18:18.590]  China.
[18:19.830 --> 18:27.130]  We assume, or the law assumes, that a foreign national is going to send everything home.
[18:28.070 --> 18:29.150]  No matter what.
[18:30.030 --> 18:37.350]  And so in those circumstances, in order for that person to work on the team, you need to get a license.
[18:37.570 --> 18:42.510]  Before they even learn anything about it, you need to get that license from the government.
[18:42.510 --> 18:46.990]  It involves all kinds of reporting, and it's an ongoing process.
[18:47.370 --> 18:47.850]  Okay?
[18:48.570 --> 18:51.590]  And that makes certain kinds of research really complicated.
[18:52.690 --> 19:02.650]  Now you have to have, as a company, a whole, like, compliance process to make sure that the people who are working on your technology are actually allowed to do so.
[19:02.790 --> 19:06.710]  If you need to have them working on it, what do you need to do to comply with this?
[19:06.830 --> 19:10.090]  It's a whole thing that employs a lot of lawyers.
[19:10.090 --> 19:11.150]  And you know what?
[19:13.410 --> 19:16.210]  Lawyers created the law to make, to sell more lawyers.
[19:16.430 --> 19:16.750]  Right?
[19:16.870 --> 19:17.830]  That's what we're all about.
[19:18.390 --> 19:21.350]  Society is a game that lawyers play with other lawyers.
[19:21.910 --> 19:23.470]  You're all welcome to hang out.
[19:24.550 --> 19:35.070]  But this is the kind of thing that, you know, there are people who make their whole lives just doing this kind of licensing stuff for big companies to make sure that they are complying with these export rules.
[19:35.110 --> 19:37.890]  Because if you recall, the penalties are big.
[19:39.530 --> 19:47.030]  And I want to, you know, loop back again to the idea that this only applies to proprietary research.
[19:47.490 --> 19:50.110]  If your company is going to get a patent on it, yeah.
[19:50.370 --> 19:52.350]  You need to be careful about deemed exports.
[19:52.490 --> 20:00.270]  If you're a university, and you're publishing everything, and it's all this free-flow utopian situation, you don't have to worry about deemed exports.
[20:01.870 --> 20:06.550]  So let's start talking about how this plays out in the modern era.
[20:06.790 --> 20:07.170]  Right?
[20:07.310 --> 20:09.930]  Everything I've talked about so far is kind of legacy stuff.
[20:09.930 --> 20:11.290]  It's been around for a while.
[20:11.630 --> 20:12.610]  Let's talk about AI.
[20:15.390 --> 20:19.470]  Early this year in January, January 15th.
[20:19.550 --> 20:20.350]  Look, I even wrote it down.
[20:21.190 --> 20:34.010]  The Biden administration, in one of its last acts in power, put in a bunch of restrictions on AI exports in the Framework for Artificial Intelligence Diffusion.
[20:34.010 --> 20:40.710]  The idea here is that AI has become, you know, it's big.
[20:40.930 --> 20:43.150]  You might not have heard of it, but I promise, it's big.
[20:43.490 --> 20:53.490]  So the country now has this interest in making sure that they stay on top, that we stay on top of the AI race.
[20:53.910 --> 20:55.230]  How do you do that?
[20:55.390 --> 20:55.710]  Right?
[20:55.710 --> 21:05.130]  We have people here creating models that are enormously complex, very difficult to replicate from scratch, but very easy to copy.
[21:06.090 --> 21:08.950]  So you want to make sure that those models don't leave the country.
[21:09.330 --> 21:09.450]  Right?
[21:10.530 --> 21:19.510]  We have model or had this proposal for model restrictions that had limits on how complex your model could be.
[21:19.910 --> 21:20.550]  Right?
[21:20.550 --> 21:20.570]  Right?
[21:21.170 --> 21:22.970]  So beyond a certain number of weights.
[21:23.190 --> 21:32.970]  And then there were clawbacks for, okay, well, if it's open-source or an equivalent complexity to something that's already open and public, then it's not restricted.
[21:33.090 --> 21:34.270]  It got very complicated.
[21:34.830 --> 21:43.390]  And that was on the books for about three months, rescinded May 13th under a new presidential administration.
[21:44.790 --> 21:54.030]  Because the new presidential administration wants a much more open approach to AI development.
[21:55.970 --> 21:57.670]  Then you've got hardware restrictions.
[21:57.990 --> 22:03.210]  And this has been a very hot topic in that I was looking at the news today to help write this slide.
[22:03.730 --> 22:06.270]  Because this has been changing very fast.
[22:07.230 --> 22:11.170]  This is a picture, by the way, of the CEO of NVIDIA looking sad.
[22:12.490 --> 22:14.390]  Hardware restrictions are another part of this.
[22:14.570 --> 22:23.030]  Because in order to make and use an AI model, certain hardware is, in some cases, practically necessary.
[22:23.430 --> 22:24.970]  NVIDIA makes a lot of that hardware.
[22:25.170 --> 22:26.690]  AMD makes a lot of that hardware.
[22:27.350 --> 22:34.390]  And so, as a result, the chips that are used for training that have been restricted.
[22:34.630 --> 22:44.210]  And this has gone into the present administration's unpredictability.
[22:47.050 --> 22:49.550]  There's been a lot of back and forth just this week.
[22:49.550 --> 22:59.970]  The latest news is that the United States is taking a 15% cut of sales of a certain chip to China by NVIDIA.
[23:00.170 --> 23:02.230]  To me, that reads as a tax.
[23:02.510 --> 23:06.590]  It reads as a tax that the President imposed unilaterally without Congress.
[23:06.870 --> 23:09.070]  I don't know if that's legal.
[23:09.310 --> 23:11.070]  I don't know what that means.
[23:11.090 --> 23:13.130]  I don't think anybody know what that means.
[23:13.130 --> 23:16.910]  You can see that I just started typing and ran off the bottom of the slide.
[23:16.910 --> 23:19.410]  And I felt like that was probably enough on the topic.
[23:20.990 --> 23:24.290]  But this is, you know, a world of madness that we live in right now.
[23:25.630 --> 23:28.330]  It's difficult to predict what's going to keep happening.
[23:28.510 --> 23:32.230]  But this all lives in the world of export restrictions, right?
[23:32.470 --> 23:34.210]  This is not a tax.
[23:34.210 --> 23:42.890]  This is some kind of limitation, or at least nobody's calling it a tax, even though it's very clearly an income tax on one specific company that's named.
[23:45.430 --> 23:45.990]  Sorry.
[23:47.450 --> 23:49.730]  I'm just going to hit next slide and move on.
[23:49.910 --> 23:53.710]  So let's talk about quantum computing.
[23:53.830 --> 23:55.290]  This is my latest thing, okay?
[23:55.390 --> 23:59.670]  So I got into this topic because of quantum computing.
[23:59.790 --> 24:01.730]  Because I started reading about it.
[24:01.730 --> 24:02.450]  It's cool stuff.
[24:02.450 --> 24:09.670]  And, oh, hey, look, in September, the Biden administration imposed new export restrictions on quantum computing.
[24:09.690 --> 24:12.350]  And that got me learning about deemed exports.
[24:12.350 --> 24:20.970]  And that got me thinking about how weird that is, that you can't even talk to a guy about a technology that you're working on because of where they were born.
[24:20.970 --> 24:23.470]  And it, you know, snowballed and here we all are.
[24:24.770 --> 24:32.350]  So quantum computing restrictions are focused on companies, or on countries in categories D-1 and D-A.
[24:32.670 --> 24:36.190]  I copied the list of D categories here.
[24:36.410 --> 24:39.590]  That's national security concerns and U.S.
[24:40.130 --> 24:41.710]  arms embargoed countries.
[24:42.090 --> 24:45.810]  And so quantum computing technologies are kind of considered both.
[24:47.830 --> 24:51.870]  Before we get too far down this topic, I hope you'll indulge me.
[24:51.870 --> 24:54.210]  Let's talk a little bit about the technology here.
[24:54.370 --> 24:55.430]  Because I think it's cool.
[24:55.530 --> 25:05.190]  But I think it helps to understand what quantum computing is used for in order to see how it fits into the idea of export controls, right?
[25:05.390 --> 25:08.070]  Because it's not just a fast computer.
[25:08.430 --> 25:11.290]  It's a very specific kind of computer.
[25:11.290 --> 25:13.230]  And I'm going to drink for a second.
[25:13.390 --> 25:16.470]  In the meantime, can somebody tell me why the U.S.
[25:16.670 --> 25:21.430]  government might be very concerned about quantum computers falling into China's hands?
[25:21.590 --> 25:21.910]  Yes.
[25:39.180 --> 25:40.060]  Great answer.
[25:40.320 --> 25:51.140]  The answer that, for the people at home, the answer was, quantum computers are very good, or potentially very good, at factoring large numbers.
[25:51.280 --> 26:00.020]  And factoring large numbers is the mathematical basis for a lot of the encryption that we've had for the past 40 to 50 years.
[26:00.540 --> 26:10.260]  There's a lot of encrypted stuff that governments around the world have just been recording and sitting on for decades, waiting to have the computing power to break it.
[26:10.540 --> 26:11.000]  Okay?
[26:12.260 --> 26:20.260]  That's a big, big reason why there's a race now to have the first workable implementation of Shor's algorithm.
[26:20.280 --> 26:22.420]  That's the factoring prime numbers one.
[26:26.120 --> 26:28.840]  There are other applications, of course, right?
[26:29.000 --> 26:32.220]  There's Grover's algorithm, which is used for searching.
[26:32.220 --> 26:33.800]  There are a handful of others.
[26:34.180 --> 26:43.900]  Generally speaking, they don't have that really nice combination of an exponential improvement over classical computers and actually being worth doing.
[26:44.480 --> 26:46.440]  That's where you get your killer app.
[26:46.520 --> 26:47.880]  And that's Shor's algorithm.
[26:49.440 --> 26:53.400]  So as a result, the government cares a lot about quantum computing.
[26:53.640 --> 26:56.620]  But quantum computing is still in its infancy, right?
[26:56.900 --> 26:58.980]  So we need a lot of research.
[26:59.300 --> 27:11.440]  There are a lot of people working on it, trying really hard to be the first past the post there to get that working implementation of Shor's algorithm that they can send and sell it to governments for one hojillion dollars.
[27:12.140 --> 27:16.060]  The second one to get there is not going to make one hojillion dollars.
[27:16.340 --> 27:18.180]  It's the first one to get there, right?
[27:18.440 --> 27:22.120]  And there's a lot of other fun stuff that you can do with quantum computers.
[27:22.340 --> 27:27.360]  There's a lot of academic interest, but it's just not where the money is coming from.
[27:27.360 --> 27:29.720]  And it's not why we have the export controls.
[27:29.860 --> 27:31.560]  It's about encryption, okay?
[27:33.600 --> 27:43.060]  So, keeping all of those motivations in mind for wanting to lock it down, we have a competing motivation, which is that we actually do want it to happen, right?
[27:43.220 --> 27:52.780]  Turns out, a lot of the very smart, skilled people who are working in quantum computing were not born in the United States, right?
[27:52.780 --> 27:59.420]  Now, countries of interest, China, Russia, Iran, North Korea.
[27:59.700 --> 28:04.440]  North Korea is not supplying a lot of first-rate quantum computing individuals.
[28:04.880 --> 28:06.820]  China is, right?
[28:07.000 --> 28:21.540]  If we were to, under the deemed export rules, say, I'm sorry, Chinese nationals cannot work for IBM or Google or Microsoft or any of the other big players on their quantum computing stuff, we just wouldn't get quantum computers, and China would.
[28:21.540 --> 28:23.680]  So, we've got this tension now.
[28:23.800 --> 28:28.720]  We have the rules in place that restrict the export.
[28:29.220 --> 28:33.760]  But when it comes to the deemed export, those rules are suspended.
[28:34.040 --> 28:40.220]  You still have to do all the same record keeping, all the same reporting, but you don't have to get the license first.
[28:40.440 --> 28:49.340]  You can bring in your friend from China or Russia or whatever to work on these technologies without getting the okay first.
[28:50.020 --> 28:53.960]  And it's kind of a, let's see how this plays out strategy.
[28:54.340 --> 29:01.280]  Now, the Trump administration has not, to the best of my knowledge, meddled with these restrictions yet.
[29:01.680 --> 29:02.280]  Who knows?
[29:02.700 --> 29:06.740]  I can't predict what's going to happen tomorrow, let alone over the next three years.
[29:06.980 --> 29:07.360]  Who knows?
[29:09.220 --> 29:12.120]  So, let's talk about what these restrictions actually cover.
[29:12.120 --> 29:15.140]  Like I said, quantum computers right now are not good.
[29:15.440 --> 29:17.420]  They cannot do things.
[29:18.280 --> 29:23.120]  The state-of-the-art quantum computers have in the low thousands of qubits.
[29:24.960 --> 29:27.060]  There are efforts...
[29:27.060 --> 29:32.660]  There are two metrics that the government uses to judge how good a quantum computer is.
[29:33.280 --> 29:41.140]  One is the number of qubits, that is physical qubits on the chip, and the other is the error rate of the CNOT gate.
[29:41.440 --> 29:52.480]  Without going too deep into it, CNOT is the operation that quantum computers use most of the time to entangle two qubits together in a quantum mechanical way.
[29:52.980 --> 29:55.880]  Happy to talk about that, but that's a very different talk.
[29:56.800 --> 30:02.320]  The short version of it is, quantum computers are not good at this.
[30:02.700 --> 30:04.980]  The error rate is high, okay?
[30:05.140 --> 30:08.220]  We are working on quantum error correction.
[30:09.080 --> 30:17.920]  And once you have that working, you can kind of throw qubits at the problem in order to get a smaller number of logical qubits, right?
[30:18.060 --> 30:19.400]  Actually working qubits.
[30:21.700 --> 30:23.940]  Or, you can just make the qubits better.
[30:24.140 --> 30:25.020]  Either way, right?
[30:25.020 --> 30:29.360]  I put in, for the sake of your edification, a couple of examples.
[30:29.640 --> 30:34.260]  We have 34 qubits or a CNOT error rate of 10 to the negative fourth.
[30:34.640 --> 30:35.620]  That's really good.
[30:36.120 --> 30:44.860]  We have 1,100 qubits, which I believe the IBM, the latest offering from IBM, has more than that many qubits.
[30:45.200 --> 30:50.480]  But I doubt they have the CNOT error rate of 6 times 10 to the negative third, right?
[30:52.000 --> 30:53.940]  Or just more than 2,000 qubits.
[30:53.940 --> 30:56.200]  Once you hit that number, you're done, okay?
[30:59.060 --> 31:05.480]  It's going to be a while before we have physical devices that by themselves meet these restrictions.
[31:07.160 --> 31:15.840]  But any time you file something with the USPTO, and this is where my personal area of expertise comes back in, there is a security review.
[31:15.840 --> 31:27.800]  And the government takes a long, hard look at this quantum computing application that you've given them, and they make the determination of whether this is something that must be kept secret or not.
[31:27.960 --> 31:37.520]  And it's kind of the first step along the way to that being deemed or considered something worthy of the export restriction.
[31:39.600 --> 31:40.560]  And that's about it.
[31:40.680 --> 31:47.200]  You know, it's 7 o'clock, or no, 8 o'clock, 8.30 on a Friday night on a dry campus.
[31:47.720 --> 31:50.660]  So I wasn't going to stress too much about, you know, filling out the time.
[31:50.660 --> 31:53.360]  I figured if we get out a few minutes early, it's not a big deal.
[31:54.000 --> 31:56.960]  But that, for your sake, means we've got lots of time for questions.
[31:57.800 --> 32:00.440]  If you'd like to know anything about any of this stuff, let me know.
[32:00.520 --> 32:03.280]  If it's within the realm of my expertise, I'll tell you.
[32:03.660 --> 32:05.020]  If it's not, I'll tell you that too.
[32:07.500 --> 32:12.300]  If you want to ask a question, I'd ask you to say it into the microphone so that the people at home can hear.
[32:14.860 --> 32:17.860]  A question about the security review that you mentioned.
[32:18.120 --> 32:20.120]  I guess it's from the Invention Secrecy Act.
[32:20.500 --> 32:32.020]  How much transparency is there with regards to, you know, how many patents have actually been classified in which categories of information, and when do they become declassified to the public?
[32:33.180 --> 32:38.660]  To what degree is the patent administration, I mean, office, transparent with those numbers?
[32:38.840 --> 32:43.640]  Yeah, I'll Google that right now, but I want you to keep looking at my email address in case you want to email me.
[32:44.700 --> 32:51.940]  So, there is an ongoing review of things that get a secrecy order on them.
[32:52.560 --> 32:58.460]  I don't know if there are good numbers available on how often things get made secret.
[32:58.460 --> 33:03.700]  There probably are, but that's not within, you know, that's not something I know off the top of my head.
[33:04.820 --> 33:05.440]  Thank you.
[33:05.700 --> 33:05.820]  Yep.
[33:06.960 --> 33:07.940]  Hey, thanks for the talk.
[33:08.420 --> 33:09.360]  I'm just curious.
[33:10.660 --> 33:16.120]  So, I know quite a few places are fully open sourcing it.
[33:16.220 --> 33:19.860]  I know IBM is, the software used to work with these devices.
[33:19.860 --> 33:25.060]  So, does that sidestep any export restrictions since it's basic research in public?
[33:25.320 --> 33:25.680]  Sure.
[33:25.940 --> 33:35.260]  So, the software that's used to program a quantum computer is not among the restrictions.
[33:35.520 --> 33:40.660]  The software, any software that you use to make a quantum computer is.
[33:41.180 --> 33:46.260]  But you're right, the IBM, I think Google also has a software package.
[33:46.840 --> 33:47.560]  Yeah, CERC.
[33:47.780 --> 33:50.560]  And, no, CERC.
[33:51.540 --> 33:54.740]  I should really know the IBM one off the top of my head, but I'm blanking on it.
[33:54.840 --> 33:54.900]  Okay.
[33:55.800 --> 33:56.200]  Ciscuit.
[33:56.420 --> 33:56.820]  Ciscuit.
[33:57.380 --> 33:57.700]  Thank you.
[33:58.420 --> 34:05.180]  But if they were, let's just say, hypothetically, they did open-source it for whatever reasons, would that sidestep it?
[34:05.300 --> 34:08.160]  Or is like quantum stuff, you know, you just can't do it no matter what?
[34:08.160 --> 34:08.460]  Yeah.
[34:08.740 --> 34:15.580]  So, if you are open sourcing it and making it, you know, public knowledge, then that would not fall under the export controls.
[34:15.800 --> 34:15.960]  Cool.
[34:16.220 --> 34:16.440]  Thank you.
[34:18.320 --> 34:26.480]  Regarding the patent secrecy, there's, I think there's about 17 or 18,000 patents right now that are secret, but there's an annual report that the government put.
[34:26.580 --> 34:28.040]  The patent office comes out with a report.
[34:28.160 --> 34:29.300]  They don't publish it, though.
[34:29.420 --> 34:32.460]  Someone has to actually request it under the Freedom of Information Act.
[34:32.600 --> 34:35.380]  And the person who used to request it each year has retired.
[34:35.380 --> 34:38.380]  So, I guess it's open to somebody to ask for it.
[34:38.500 --> 34:42.280]  But my question is this, you talked about patent exporting.
[34:42.500 --> 34:45.080]  You talked about technology export limitations.
[34:45.560 --> 34:48.960]  Are there any technologies which are prohibited from import?
[34:49.240 --> 34:52.420]  And I guess I can think of, like, cell phone jammers, maybe.
[34:52.600 --> 34:57.700]  But is there any technologies, is that a thing that prevent technology from coming into the country?
[34:57.700 --> 34:59.100]  That is so a thing.
[34:59.300 --> 35:00.600]  That is intensely a thing.
[35:00.700 --> 35:04.260]  Not so much in, like, in the same sense of technology.
[35:04.500 --> 35:09.120]  But there are big restrictions on what comes across the border.
[35:09.340 --> 35:11.980]  And it's usually more economic-based, right?
[35:11.980 --> 35:21.680]  We don't want this sort of thing coming in to, you know, interfere with our domestic industries.
[35:21.680 --> 35:28.080]  Like, there's one thing in particular, which I hesitate, you know, to broadcast.
[35:28.360 --> 35:36.300]  But the, let's say, reproductive matter for certain livestock is heavily controlled.
[35:36.560 --> 35:37.840]  I wouldn't call it a technology.
[35:37.900 --> 35:46.620]  But it's certainly on the sort of, like, restrictions of import and export that is similar to what we're talking about here.
[35:48.960 --> 35:53.180]  Hey, I work at IBM Quantum, actually, and really enjoyed this talk.
[35:53.340 --> 35:53.720]  Oh, thank you.
[35:53.920 --> 36:04.720]  My question is, why do companies like IBM have to geo-restrict our access to things like our cloud service and even our documentation we geo-restrict?
[36:08.160 --> 36:08.760]  Uh...
[36:09.480 --> 36:10.800]  I mean, I don't know.
[36:11.700 --> 36:12.200]  Are we...
[36:12.200 --> 36:14.760]  Is this a question about exports?
[36:14.760 --> 36:20.500]  Yeah, like, my understanding is we do it for export reasons, that we have to do it for compliance.
[36:20.640 --> 36:24.440]  I'm wondering with what you're saying with export controls, why would we...
[36:24.440 --> 36:34.800]  It sounded like Quantum is more loose with export controls and other technologies, so why do we still have to restrict access to certain countries based off of geography?
[36:34.800 --> 36:40.880]  Like, if you log in with an IP address from Iran or Egypt, for example, you get blocked.
[36:40.880 --> 36:42.420]  I mean, it might be...
[36:42.420 --> 36:44.640]  So, when you say we, who is we here?
[36:44.840 --> 36:45.280]  Help me, give me some...
[36:45.280 --> 36:45.880]  IBM Quantum.
[36:46.140 --> 36:46.780]  Okay, yeah.
[36:46.920 --> 36:51.020]  And I imagine Google and others are doing it as well, but I know at least IBM Quantum is.
[36:51.200 --> 36:51.880]  Okay, awesome.
[36:52.540 --> 36:53.020]  Um...
[36:54.020 --> 37:09.880]  I am going to guess that it's out of an abundance of caution, that they could stop and figure out exactly who on what team is working on stuff that could potentially be considered an export, but that's a lot of effort.
[37:16.410 --> 37:17.010]  Um...
[37:17.010 --> 37:20.710]  I mean, that is one of the...
[37:21.210 --> 37:23.730]  So, that's one of the governing bodies that...
[37:23.730 --> 37:25.930]  or bodies of rules that deals with it.
[37:26.290 --> 37:27.850]  ITAR deals more with...
[37:27.850 --> 37:30.030]  The question was, could it be ITAR?
[37:30.770 --> 37:40.790]  ITAR is one of the two bodies of regulations that deals mostly with export controls, and ITAR is more about the weapons side of things.
[37:41.430 --> 37:41.970]  So...
[37:41.970 --> 37:42.510]  Maybe?
[37:42.910 --> 37:43.930]  Maybe, right?
[37:44.010 --> 37:44.510]  I don't know...
[37:44.510 --> 37:53.770]  I don't know why IBM makes their decisions and how they, you know, pull those policies, but it's entirely possible that they do it because they are worried about this specific issue.
[37:54.190 --> 37:54.750]  Thank you.
[37:56.650 --> 37:57.190]  Hi.
[37:57.430 --> 37:57.970]  Thanks for the talk.
[37:58.750 --> 37:59.270]  Um...
[37:59.270 --> 38:12.810]  From your sense of the legal story around model export restrictions, if we did see those come back in some reduced form, do you have any sense of how those would be impacted by developments in open-source models getting released?
[38:13.110 --> 38:13.510]  Yeah.
[38:13.790 --> 38:18.550]  So, you know, for a period of three months, I was thinking, oh man, this is going to be really interesting.
[38:18.690 --> 38:20.650]  I wonder how this is going to impact, you know?
[38:20.850 --> 38:36.930]  Because in that period of time, China started coming out with some really powerful models of their own, and built into those regulations was the idea that like, all right, well, if there's kind of an equivalent model out there, then you don't have to worry about exporting the U.S.-made models.
[38:37.290 --> 38:43.570]  Uh, so there's like, kind of built, a built-in arms race in how those regulations were written.
[38:44.590 --> 39:00.650]  Um, I don't know how you enforce that without somebody whose job it is to keep, you know, just sift through the news, uh, worldwide of every AI model that comes out and kind of like, move the ponies on the board to see who's winning, right?
[39:00.730 --> 39:01.270]  Like, I don't know.
[39:01.610 --> 39:07.610]  Um, I don't foresee them coming back in the next three years, put it that way.
[39:07.930 --> 39:18.030]  Uh, just because the present administration has a very aggressive policy about, uh, lessening the restrictions on developing AI.
[39:18.830 --> 39:19.470]  Thank you.
[39:20.810 --> 39:21.330]  Hi.
[39:21.750 --> 39:22.070]  Hi.
[39:22.190 --> 39:22.850]  Thank you for the talk.
[39:23.050 --> 39:24.470]  Um, I had a pesky hypo for you.
[39:24.710 --> 39:26.270]  Oh, that's kind of hypo.
[39:26.270 --> 39:35.690]  So I'm curious, uh, whether we're talking about, um, quantum computing technology or stepping back to the Nvidia example of AI chips used for useful for training.
[39:36.670 --> 39:44.530]  Hypothetically, if I'm a vendor, all of my hardware is based in the U.S., but I sell client tenant based access to that hardware through some sort of virtual instance.
[39:44.870 --> 39:45.990]  Am I exporting?
[39:46.270 --> 39:49.110]  So services can count as an export.
[39:49.710 --> 39:52.410]  Um, obviously it's a, it's kind of fuzzy, right?
[39:52.410 --> 39:57.430]  Like, are you exporting the technology?
[39:58.250 --> 40:09.590]  Um, I think that if you're just, so for example, IBM's quantum computing business model right now includes anybody can create an account.
[40:09.790 --> 40:10.590]  Maybe not anybody.
[40:10.590 --> 40:11.070]  I don't know.
[40:11.170 --> 40:23.770]  Maybe they've got geographic restrictions on, you know, who's allowed to sign up for an account, but it's available to the public where anybody can get time on an actual quantum computer and run their software on it for research purposes, for whatever purposes.
[40:23.770 --> 40:38.110]  Um, as it stands, I don't think that would count as an export because the technology for how to create that computer hasn't left the United States.
[40:38.310 --> 40:38.430]  Right.
[40:38.650 --> 40:40.990]  But that doesn't mean that there aren't other world, right?
[40:41.090 --> 41:04.310]  So like, for example, if you had a service where you transmitted your stuff to a U.S. based computer computer and then encrypted it and then sent it back, I could see that violating encryption restrictions because now the service of encrypting with whatever strong encryption is available to other
[41:04.310 --> 41:05.110]  people, right?
[41:05.250 --> 41:07.950]  To, to whatever nation you're worried about.
[41:09.310 --> 41:09.870]  Interesting.
[41:10.010 --> 41:10.650]  No, that, that makes sense.
[41:10.790 --> 41:20.310]  I'm thinking of the AI training model where, you know, there's a sort of tangible difference between I'm lending you GPUs to run this on versus I'm storing your trained model and then sending it back to you.
[41:20.470 --> 41:20.870]  Yeah.
[41:21.010 --> 41:25.490]  I would see like if you were training a model for someone and then sending them the model back.
[41:25.690 --> 41:25.810]  Right.
[41:25.990 --> 41:28.150]  You are then exporting that model for sure.
[41:28.610 --> 41:29.030]  Thank you.
[41:30.390 --> 41:33.470]  Um, there's one loophole I'm, I'm aware of.
[41:33.510 --> 41:34.590]  I don't know if it's still in effect.
[41:34.610 --> 41:35.770]  It was in the 2010s.
[41:35.850 --> 41:42.910]  I know is if you were a U.S. based financial services company, you may use encryption algorithms that are restricted.
[41:43.290 --> 41:49.310]  Uh, if you were doing business with the country on the other end, even if you're not allowed to sell the encryption to that country.
[41:50.310 --> 41:56.190]  So in that case, the, the other country already has the encryption, obviously.
[41:56.630 --> 41:59.590]  Well, because you know, the idea is I'm, I'm a bank.
[41:59.750 --> 42:08.390]  I'm using an encrypted now encryption algorithm is considered non exportable, but I can use it on the computers that I run in that country.
[42:08.550 --> 42:10.970]  I see because you are a U.S. bank.
[42:10.970 --> 42:16.110]  You can run your U.S. encryption in outer Slobovia.
[42:17.090 --> 42:23.450]  And even though outer Slobovia is an enemy of our nation, you are able to communicate with your office there.
[42:23.450 --> 42:24.230]  I get it.
[42:24.370 --> 42:26.130]  Except when it comes to banking, we're all Swiss.
[42:27.610 --> 42:32.130]  Uh, one, one side effect of that, that somebody had mentioned about the cloud restrictions.
[42:32.270 --> 42:46.250]  Uh, I know of at least two large banks that run private clouds for that reason, because then they can have a private cloud in Slobovia, uh, that has all of the necessary, uh, encryption that they need to run.
[42:46.410 --> 42:46.630]  Right.
[42:47.030 --> 42:48.010]  Uh, along those lines.
[42:48.550 --> 42:56.610]  But, uh, that, uh, also another one is post quantum encryption, by the way, I think is probably going to be, if it isn't already, it's going to be restricted.
[42:56.610 --> 42:57.150]  Yeah.
[42:57.650 --> 43:08.970]  I mean, post, post quantum encryption, people, people overstate the impact of quantum computers on encryption, right?
[43:09.510 --> 43:10.110]  Yes.
[43:10.550 --> 43:12.050]  RSA is done.
[43:13.030 --> 43:29.590]  But, the modern encryption that we all use, AES, it's not post quantum encryption in the same way, but the speed ups of doing it on a quantum computer are not big enough to say that it's broken in the same way that RSA is once you've got, like, uh, uh,
[43:29.930 --> 43:31.490]  an implementation of Shor's algorithm.
[43:31.490 --> 43:40.690]  Um, we're going to have, like, whatever we're transmitting right now under modern encryption is going to stay secure for decades.
[43:41.030 --> 43:54.330]  It's going to take a long time before quantum computers get to be any kind of viable, let alone viable, you know, strong enough to deal with today's strong encryption.
[43:54.770 --> 43:59.950]  So, yes, post quantum encryption is an active field of research.
[44:00.510 --> 44:03.210]  But I wouldn't stress about what we're transmitting today either.
[44:03.510 --> 44:03.650]  Yeah.
[44:03.750 --> 44:09.210]  I mean, the way it's being treated by the large banks that I do business with is that they sort of have a plan.
[44:09.390 --> 44:10.350]  It's like Y2K.
[44:10.690 --> 44:14.490]  At a certain point, they're going to pivot to post quantum encryption.
[44:14.730 --> 44:18.010]  So, it's, it's, there's a plan that will happen at some point in the future.
[44:18.110 --> 44:18.130]  Yeah.
[44:18.130 --> 44:18.790]  And they should, right?
[44:18.870 --> 44:19.190]  That's good.
[44:19.270 --> 44:20.350]  You want to stay ahead of the curve.
[44:20.570 --> 44:20.850]  Right.
[44:21.090 --> 44:23.830]  But it's like, it's, it's not, you know, it's like, it's on a timeline.
[44:24.050 --> 44:26.770]  It's not like it, it's a no, there's no race to do it.
[44:26.970 --> 44:27.050]  Yeah.
[44:27.910 --> 44:28.270]  Thanks.
[44:29.490 --> 44:33.410]  Uh, so I had a question about the extent of export controls.
[44:33.630 --> 44:33.830]  Yes.
[44:33.830 --> 44:43.890]  I know you said that it was, it pertained to services, physical devices and technology, but does it also apply to people who know that technology as well?
[44:45.170 --> 44:45.570]  Yes.
[44:46.430 --> 44:46.830]  Yes.
[44:46.990 --> 44:47.270]  For sure.
[44:47.410 --> 44:48.510]  That's the deemed export.
[44:49.250 --> 44:51.030]  Um, we're just, yes, go ahead.
[44:51.290 --> 44:51.450]  Yeah.
[44:51.550 --> 44:55.230]  So like, let's say I, I was working on something that has an export control.
[44:55.390 --> 45:02.610]  Then if I wanted to leave the country to one of those, uh, countries, then I would have to get a license or be restricted.
[45:02.610 --> 45:04.610]  So it's not about the travel.
[45:04.750 --> 45:08.650]  It's about the, the kind of national identity.
[45:08.890 --> 45:17.190]  The rule relates to where the person was last a citizen or last a resident for some period of time.
[45:17.390 --> 45:22.290]  Um, so it's not so much about leaving the country that triggers it.
[45:22.290 --> 45:27.350]  It's about where you were, where you came from before you came to the country.
[45:28.370 --> 45:36.010]  The, you know, one, you know, it's, it's assumed that as soon as the person learns this stuff, that it's already been exported.
[45:36.570 --> 45:36.850]  Right.
[45:37.030 --> 45:38.990]  You don't have to leave in order to export something.
[45:39.070 --> 45:40.190]  You can transmit it on the Internet.
[45:40.190 --> 45:41.010]  You can write a letter.
[45:41.770 --> 45:43.070]  So it's, as soon as that.
[45:43.170 --> 45:46.750]  And so that the restriction really tries to keep that from happening.
[45:47.630 --> 45:54.690]  But, you know, keep in mind, academic research, fundamental research, doesn't fall under these restrictions.
[45:54.970 --> 46:06.290]  So somebody who is learning something while they're here, because they're studying, they're doing, you know, whatever it is they're doing and they go home, they haven't violated the export restrictions because they're doing fundamental research here.
[46:07.110 --> 46:07.630]  Okay.
[46:07.750 --> 46:08.030]  Thank you.
[46:09.530 --> 46:10.290]  Welcome back.
[46:10.630 --> 46:10.930]  Thank you.
[46:11.170 --> 46:11.770]  Second question.
[46:12.350 --> 46:16.530]  Um, this generates, uh, is related to patent law more generally.
[46:16.810 --> 46:19.990]  Um, the concept of disclosure to the public.
[46:19.990 --> 46:35.570]  So the period of time between I've conceived of a new invention and I haven't yet, uh, filed a patent or provisional patent, you know, my understanding is if you disclose that invention to the public, you lose the protection, uh, you know, of the patent.
[46:35.730 --> 46:39.810]  But my question is what really, what is a technical definition for disclosure?
[46:40.130 --> 46:43.890]  Does it have to be enough information for somebody trained in the art to replicate the idea?
[46:43.890 --> 46:46.910]  Or, or, you know, what, what level of...
[46:47.550 --> 46:47.930]  Okay.
[46:48.110 --> 46:49.630]  So first of all, I want to correct.
[46:49.890 --> 46:50.110]  Okay.
[46:50.190 --> 46:50.930]  I want to correct you there.
[46:51.150 --> 46:56.270]  Because let me remind everybody, I am a patent attorney and that's my email address in case you've got patents.
[46:56.790 --> 47:04.970]  Um, if you disclose something between the date of filing and, you know, when it becomes published, you haven't lost anything.
[47:05.170 --> 47:06.670]  Your filing date locks it in.
[47:06.810 --> 47:11.370]  If you disclose it before you file your application, that's when things start to get hinky.
[47:11.570 --> 47:16.990]  And there are two different kinds of ways in which disclosure starts to be an issue.
[47:16.990 --> 47:26.670]  There's disclosure in terms of when you, when you put something out in the world.
[47:26.930 --> 47:31.630]  So it starts to become like prior art in that sense.
[47:31.770 --> 47:38.610]  And there's disclosure in the sense of, is this whatever document this is and enabling disclosure.
[47:38.870 --> 47:39.110]  Right?
[47:39.290 --> 47:48.350]  And so when you're talking about a written, you know, scientific article or whatever, that has to be enabling.
[47:48.350 --> 47:51.110]  It has to explain how to do the invention.
[47:51.450 --> 47:54.790]  Whatever piece of it is that the patent office is relying on.
[47:54.930 --> 47:56.690]  It has to actually explain what's going on.
[47:56.790 --> 47:59.550]  Let me give an invented example.
[47:59.890 --> 47:59.990]  Sure.
[48:01.290 --> 48:02.710]  Here's a better mousetrap.
[48:03.050 --> 48:08.350]  And we zap the mouse with a laser instead of a snap, you know.
[48:08.630 --> 48:18.250]  So if I don't describe, well, what wavelength was the laser and what energy and the physical, well, you know, all the details, but here's, here's generally the conception.
[48:18.630 --> 48:18.790]  Yeah.
[48:18.930 --> 48:20.210]  Is that a disclosure or not?
[48:20.470 --> 48:23.110]  No, you've put on your blog, look at my cool laser trap.
[48:23.370 --> 48:25.850]  I can, it can trap laser mice, right?
[48:25.950 --> 48:28.350]  But you haven't explained everything about how it works.
[48:28.550 --> 48:28.630]  Right.
[48:28.930 --> 48:32.070]  That's a, that's a very fact sensitive question.
[48:32.070 --> 48:33.690]  How much have you actually given out?
[48:33.870 --> 48:36.490]  Is the only idea that it's a laser on a mousetrap?
[48:36.650 --> 48:38.630]  If so, then yeah, you kind of just disclosed it.
[48:38.630 --> 48:42.070]  And if there's like some deeper technology, maybe not.
[48:42.490 --> 48:42.930]  Right.
[48:43.090 --> 48:45.910]  So it gets down to what you've actually told people.
[48:46.150 --> 48:46.350]  Thanks.
[48:47.450 --> 48:47.810]  All right.
[48:48.090 --> 48:51.010]  And we're going to make this the last one because we are running up on time.
[48:51.590 --> 48:52.030]  Okay.
[48:52.250 --> 48:52.590]  That's good.
[48:52.670 --> 48:53.910]  Cause it's like barely a question.
[48:54.330 --> 48:57.010]  I read a book, Apple and China recently.
[48:57.010 --> 48:57.410]  Okay.
[48:57.670 --> 49:09.590]  And just about how maybe inadvertently there's just this, almost a Marshall's plan or two worth of, you know, financial benefit that went in that direction, you know, the information, et cetera.
[49:10.930 --> 49:14.390]  Isn't it that inevitable anyway with the way we work?
[49:14.390 --> 49:17.170]  And then will any of these sort of controls work?
[49:17.610 --> 49:22.250]  Or is it only just to kind of give us just the slight edge and then we're all to that next point.
[49:22.370 --> 49:23.610]  And then there's just another edge.
[49:23.710 --> 49:24.630]  Is it constantly that?
[49:24.870 --> 49:25.790]  No, that's a great question.
[49:25.790 --> 49:29.310]  And I think you've, you've got the, you've got the truth of the matter, right?
[49:29.430 --> 49:34.150]  All of these controls, we are, you know, except when it comes to like physical goods, right?
[49:34.190 --> 49:39.950]  If we're talking about a physical device, okay, if I don't sell you this physical thing, you're not going to have it.
[49:40.070 --> 49:42.050]  If you can't make your own, you're out of luck.
[49:42.430 --> 49:45.730]  When it comes to inchoate forms of technology, right?
[49:45.810 --> 49:47.150]  The know-how, the software.
[49:47.890 --> 49:48.170]  Yeah.
[49:48.710 --> 49:53.330]  You're, you're trying to hold back the tide and that's just an ongoing process.
[49:53.330 --> 49:55.050]  You save yourself a little bit of time.
[49:55.050 --> 50:00.010]  And then you update your rules to reflect the current technology and what is actually going on right now, right?
[50:00.150 --> 50:03.350]  Like this, this AI stuff literally is happening right now.
[50:03.530 --> 50:04.690]  It happened this week.
[50:04.690 --> 50:09.870]  So like it's, it's an ongoing process as we've developed new technologies.
[50:09.870 --> 50:12.010]  We recognize the risks of new things.
[50:12.170 --> 50:14.510]  We realize that old things aren't a big deal anymore.
[50:14.510 --> 50:16.510]  And it's a, it's an involving body.
[50:18.210 --> 50:20.030]  And with that, I am going to wrap it.
[50:20.170 --> 50:21.410]  Thank you all so much for coming.
[50:21.410 --> 50:21.610]  Good morning.