概要
In this conversation, Tim and Chris examine how artificial intelligence is reshaping the landscape of intellectual property, particularly in patent law and trade secret strategy. They explore the concept of FOSITA (person of ordinary skill in the art) and how AI’s growing capabilities challenge traditional notions of inventorship, obviousness, and enablement. As AI increasingly assists or even drives innovation, they highlight the critical importance of clearly documenting the human role in the inventive process to safeguard patent validity. The discussion also addresses the rising risk of trade secrets becoming readily ascertainable in an AI-driven world, emphasizing the need for a more integrated, system-level approach to defining and protecting proprietary knowledge.
Takeaways
- AI is making it harder to define who the true inventor is.
- Including a human name on a patent may not be enough to ensure validity.
- The standard for what’s “obvious” in patent law is rising due to AI tools.
- Documenting the human contribution is critical for defensible patents.
- AI may make trade secrets easier to reverse-engineer or infer.
- Discrete trade secrets are more vulnerable—system-level knowledge is harder to replicate.
- Companies need to rethink how they structure and protect IP in an AI-driven world.
- Both patents and trade secrets are under pressure from rapid AI advances.
Tim (00:01.053)
And we’re back. Number 15. Chris.
Chris (00:02.284)
Hey Tim, good to see you.
Tim (00:05.791)
Likewise, man. How’s life?
Chris (00:09.794)
Things are good, lots of interesting things are going on. Lots to talk about.
Tim (00:14.033)
Yeah, so, what’s a faux sida?
Chris (00:20.194)
Yeah, you love pronouncing the acronym. So there’s this idea in patent law of a person of ordinary skill in the art. And it’s a long, long name. So people either do the acronym or they’ll just say a skilled artisan. But there is no person, like there’s no one person you can point to. It’s actually a fictitious legal creation of
an average person in the field. And depending on the field, that might be just a normal person, or it might be a PhD, a postdoc, a Nobel Prize winner. It could be someone super educated and super skilled. It all depends on which field you’re talking about.
Tim (01:12.617)
I guess the question we’re gonna dig at today is like, could it be AI?
Chris (01:18.604)
Maybe so, that’s actually a great question for us to talk about. Of course, AI is not a person. So the P, the first part of your acronym is person. And at least in the patent field, lately they’re saying person is a human, like an actual human being like you and me. So AI for patent purposes can’t be an inventor. At least that’s the way.
Tim (01:22.891)
Ugh.
Tim (01:37.835)
That’s right.
Chris (01:47.198)
The US and the UK and other systems are interpreting inventorship laws. But it’s an interesting field.
Tim (01:55.019)
OK. Yeah, and I think even just finding the right framing, and sorry to start off with the acronym that is FOSITA, P-H-O-S-I-T-A. Oh, gosh, that’s a mouthful. Anyway, and that really sort of comes into play when you talk about whether or not a patent’s obvious, right? If you try to reject, a patent invalidated based on obviousness claim.
Chris (02:23.576)
Well, it actually comes into play a couple different times. So why do you need to create this fictitious legal person? So one is obviousness, which we’ll talk about a lot. The other is enablement. How much do you have to teach and guide someone in a patent application depends on who’s reading it. So the more skilled and better an expert it is, the less guidance they need.
to make and use your claimed invention. So this person of ordinary skill in the art, skilled artisan, actually comes into play a couple of different ways. And it’s either a good thing or a bad thing depending on if you want the patent to be valid or if you’re trying to knock out the patent.
Tim (03:12.683)
That’s right. OK, so where this got started in the last 48 hours of our life is I put out this LinkedIn post, soliciting all of our friends to point us to anything written about this topic of the intersection of AI and patents, in particular when it comes to obviousness or non-obviousness.
and the topic of trade secrets and that is, is a sort of a different standard, but, but, but similar in a way readily ascertainable. so anyway, we’ve got some really great, a few really great, sort of followups on that thread. There’s a piece that, I’ll just give a little shout out to Robert Plotkin, really great piece that he.
wrote somewhat recently, January, 2025. He’s a patent attorney, does a lot in software. And this piece that he published on IP Watchdog is really great. And it really sort of gets at the heart of the question I was asking, which is essentially, if you can envision a world, well, even
today’s world, but let’s just like project out a little bit further where these AI systems are incredibly capable at predicting things and having the answer. And I think this example that we talk about, or he talks about in the article is, you know, some chemistry, right? You know, a new molecule for whatever, or, you know, a new design for a semiconductor.
can imagine now that this person having ordinary skills in the art has access to these AI systems, which can kind of give you an answer sort of straight away, right? So it’s not like you had to toil for five or 10 years to figure out that answer. You just went over to your computer, typed in some inputs and there’s your answer, right? So there’s your new molecule and it should work for all these reasons, you know, go make it.
Tim (05:37.939)
I guess the somewhat cynical is not quite the right word, but I’ll just say cynical view is, okay, if the entirety of human knowledge is going to be readily accessible, if new and interesting capabilities are going to be a few clicks away or answers to problems are just a few clicks away, like where in the heck does that leave the patent system?
And then we’ll get into trade secrets in a bit, but I guess I postulate, like, isn’t everything going to be obvious in a way or meet that obviousness standard simply because skilled artisans are going to know, well, I use this model to do this, right? And there’s the answer.
Chris (06:29.194)
Yeah, and obviousness, it’s one of those words that give patent lawyers headaches or conniptions because it doesn’t mean what it means to average normal people. So in the patent world, it’s not obvious to try. Like everything’s kind of obvious to try, but obvious means it’s obvious that it will work with some very high percent likelihood.
Tim (06:36.907)
I could see it in your eyes.
Chris (06:59.368)
And AI, I always say AI is a tool. Like fundamentally AI is another tool that happens to be incredibly powerful. But it makes that prediction of whether something will work or not much more predictable. So you’re right. Like the ability to have someone say, or a patent office say, your invention that you’re trying to patent is obvious.
becomes a much more compelling argument. And you as a patent seeker, patent applicant, you’re gonna have to jump over a higher bar than you used to because the tools are suddenly much more powerful. So it really, it will change the landscape and make it more difficult for a patent applicant.
Tim (07:53.641)
Yeah. And I’ve read Robert’s IP watchdog piece a couple of times now. It’s really good. It’s optimistic for sure that the patent system will still be standing. It’ll find a way to adapt to this new world order. think one of the super interesting things that’s probably happening in real time right now, and nobody’s quite
sort of put their finger on it, is that you have machines that are inventing. I mean, let’s be serious. The machines are totally inventing, okay? And then the human steps in, does some analysis, and that sort of collective becomes the invention. And that collective goes into the patent, okay? So I’m being maybe a little crass, but that’s kind of probably what happens in a lot of today’s invention. Okay.
You can see a world like forming sort of very quickly where you’re going to have people trying to invalidate patents left and right because they’re going to say, the machine invented that. You didn’t invent that. And so isn’t it, aren’t we sort of staring down the barrel of this? Like, well, you’re going have to sort of, you’re going to have to be in a position as the inventor to show where the invention was.
but where the human step was.
Chris (09:26.102)
Yeah, that’s totally right. And it’s kind of an extension of a normal way of invalidating a patent, which is to attack the inventorship. And people have been doing that for a long time where there’s either a person missing who should be on there, or what more often happens is a person’s on there who should not be. You add the PI of the lab to every single patent.
Tim (09:36.511)
That’s right. Yeah. That’s my favorite. Yeah.
Tim (09:44.928)
Yep.
Tim (09:50.592)
Lovely.
Chris (09:54.546)
You add a technician because they worked really hard, but they didn’t actually invent anything. But they spent 80 hours a week in the lab. So it’s very normal thing to do early in a patent litigation is the defendant will scrutinize the list of inventors. And if there’s anything wrong with that list, that’s a grounds to invalidate the patent.
Tim (09:58.409)
Yeah.
Chris (10:21.006)
I agree with you. think there’s a 99 % chance that people right now, today, are looking at the laws and saying, well, if AI by itself invents something, I can’t patent that. So why not just add a human? Basically, you know, skirt the problem or avoid the problem by just adding Tim or Chris or whoever, even though arguably,
Tim (10:45.727)
Yeah.
Chris (10:50.542)
you didn’t actually invent anything. And it will probably make it through the patent office because the patent examiner won’t have that information available and really has no mechanism to figure out did Tim and Chris invent anything. They just kind of mostly take it on faith or just rely on the applicant to get it right. And it won’t be until there’s litigation that it…
Tim (10:57.823)
Right. Yep.
Chris (11:19.212)
you it comes out that you and I actually did nothing, but we were listed there. Therefore, the inventorship is wrong. Therefore, the patent’s invalid. Therefore, the defendant wins. So it’s kind of, it’s not a whole new way of invalidating a patent, but it is a different wrinkle where, you know, I’m guaranteed people are adding a human inventor, well, inventor in quotes, today.
Like I’m sure it’s happening already.
Tim (11:49.247)
Yeah. Yeah. Yeah. No, my, my favorite version of that, by the way, is like, you know, you’d see these, you know, you’d see somebody’s patent, right? A company, you know, or competitor and it’d be like, yeah, two really obvious inventors. Yeah. Those, those people are clear subject matter experts and you know of them. And then you see this like third name. You’re like, this guy’s like, it’s like an investor, you know, like I didn’t invent anything. What is it going on here? Those are always my favorite. You know, when somebody that has like
Clearly no technical skill in that art gets listed.
Chris (12:21.678)
Well, or the other option is, yeah, every single of the thousand patents in the company have the CEO on it who’s out busy raising funds and schmoozing investors. It’s, yeah, it’s just hard to believe. And those also happen all the time.
Tim (12:30.749)
Yeah, exactly. That’s actually better. Right. That’s right.
So.
The back, so.
Tim (12:48.859)
This is like, think really the first time. So look, you and I’ve been talking a lot about the convergence, convergence, not the right word. We’ve been talking a lot about the sort of the intersection of patents and trade secrets, right? Where does, where’s the patent end and the trade secret begin? And so what it just struck me as I was starting to think through this stuff, that’s like, wow, this is really
this is really the first time that I could really like clearly see that while that, you know, we preach all day and every day, look, you got to write your trade secrets down, right? If you really want to be able to enforce them confidently, you got to know what they are, right? And lots of reasons to do that. This is almost like the, this is like in a weird way, an extension of that or the converse or something where you’re saying, look, in terms of the human step of whatever invention was, and if you’re,
read Plotkin’s piece, it’s pretty good. It’s sort of saying, yeah, like, hey, I got all this data out of the machine. I went off and I thought about it. I saw this pattern. I saw that pattern. Good, good, good. I went and, whatever, told the machine to go do some more stuff, did some more analysis. There you go. That whole package. That’s the innovation, right? Now, what part of that are you going to patent? Well, that’s going to depend, right? But certainly that human step piece of it is critical. And it does feel a little bit like
writing that down and making it clear that that was a human step is probably like super important right now. And it’s not clear anybody really doing that, you know?
Chris (14:21.642)
It will be if, yeah, if you have any plans or risk of ending up in patent litigation, that needs to be well documented in advance of, you know, which, what role did the human being actually play in inventing whatever it is you’re claiming. And if that’s vague and fuzzy, then the defendant, you know, the non patent holder, the patent infringer.
alleged, is going to say, you know, that whole thing was AI and you just added a human to comply with the requirement. But you you’re, you’re, don’t deserve your patent. It’s invalid. You know, you lose. So yeah, I think the, the better, the better you document it, the better a chance you’ll have of surviving at least that attack.
Tim (15:08.159)
Feels, yeah.
Tim (15:17.897)
Yeah. It feels.
Chris (15:19.852)
And it’s really like, there’s, there’s like three categories on the spectrum. So one is the old fashioned purely human, you know, the inventor toiling away or taking a shower comes up with a brilliant idea. And then the other extreme is purely AI, which consistently courts are saying, yeah, AI cannot be an inventor by themselves. And then it’s really this middle ground that very few people are talking about, which is why.
Robert’s article is so refreshing, is humans using AI as a tool. And what are the challenges and opportunities of that situation because AI is so powerful? So that’s sort of the middle ground, which I think will be more and more important. And as usual, the better you document how things were developed and how you came up with ideas,
the better off you’ll be.
Tim (16:21.055)
Yeah, also it can’t help but think that…
you know, it’s much like sort of now being able to see AI in writing. You’re like, yeah, that’s AI, you know? I just wonder what that looks like as well from a patent examiner capability, you know, where they’re able to actually have their own models that help them understand, was there really a human step in here? That’s like super-
Chris (16:55.532)
Yeah, and the patent offices are thinking of ways and gradually deploying AI themselves to make them more effective, more efficient. So it’s being used on both sides of the patent offices as well.
Tim (17:12.491)
Yeah. And I think the, again, back to my like slightly cynical side would be, yeah. So we sort of are sitting here today thinking, the invention. So let’s just say, yeah, I saw this pattern. I went to a machine, you know, came up with something, added my little human touch. Fine. Um, but guys, it just.
feels to me like whatever that process I just said, right? I saw this molecule, I went through this process out of, you boom. that, I just see a day in the not too distant future where that’s just done already. You know what I’m saying? Like it’s almost like the corpus of human knowledge is just gonna go on some weird exponential. And then it’s like kind of a prior art issue, you know?
It’s just, it’s already out there, you know? Like, you know, name your favorite, like encyclopedia or Wikipedia or whatever, you know, whatever, you know, data source you want. It’s just gonna already be there. I don’t know. Maybe I’m getting a little too…
Chris (18:24.238)
Well, no, I think that that will happen. you know, if you invent a new drug or a new chemical compound for anything, you know, when you write a patent application, you’ll usually make some variations. Like you might have five or 10, you know, similar, but different molecules. But using AI or using computers, you can generate 10,000 variants. And all of that can be disclosed or published or
Tim (18:43.669)
Right.
Chris (18:54.144)
any, you know, put in the public domain somehow. And now suddenly, instead of having five molecules as prior art, you have 10,000 or a hundred thousand molecules. So just the amount of information that’s out there that’s reachable through Google or search engines or, you know, CAS online is staggering. So, you know, all of these things make it just that much harder to get a.
patent that survives attack. And we always talk about the patent system being like, it’s not broken or bad or useless. It’s just less shiny than it used to be. Like the patent system used to be incredibly shiny. It was great for anything that ails you. But what’s happening now is some of that shine is coming off and the patent system is still useful, but it’s less useful and
Tim (19:25.536)
Yeah.
Chris (19:52.394)
again, less shiny than we used to.
Tim (19:55.115)
All right, so let’s go into readily ascertainable for a few minutes because I want to be fair. If we’re going to pick on patents, we’ve to pick on trade secrets a little bit here too. again, back to that example is I go to the machine, I train it to do something. I would not patent that. That training of how this machine
Yeah, it was trained, what it’s doing, why it’s doing it. That’s all sort of, you know, that’s all trade secret sort of territory. You know, the minute you go off into, you know, human intervention, something that’s not, you know, super easy to reverse engineer. Yeah, that’s where you want to put your patent cap on, right? But much in the same way, if I say, Hey, this is my trade secret, right? This machine.
training protocol, cetera, et cetera, right? Whatever it is. What do you think?
What do you think happens in that same world order where you’ve got this infinite amount of human knowledge or infinite amount of capability with computers where you can just say, hey, go figure out how their algorithm works. Like in other words, nobody can figure out quite figure out the TikTok algorithm, but maybe give it a few more years and the machine will figure it out. And so that trade secret that was the TikTok algorithm,
maybe isn’t really a trade secret anymore because now the machine can predict it. How do you react to that? I’m still freaked out.
Chris (21:44.366)
Well, and readily ascertainable. Yeah, that’s a long question with an equally long answer. The there’s really two parts to readily ascertainable. So one is just how easy or how difficult was it to ascertain? So like if you can hop on Google or, you know, using the AI tool of your choice quickly and easily figure it out. It’s readily ascertainable.
Tim (21:48.467)
you
Chris (22:12.558)
So that’s kind of the ease factor. The other one is kind of a fairness or acting honorably factor. So like, if you just sit down at your laptop and type something in Google, you haven’t done anything that feels wrong. Like you haven’t acted in a way that’s not honorable. On the other hand, you pretend to be someone that you’re not,
Tim (22:33.653)
Yeah. Yep.
Chris (22:42.638)
to get a log in to a system that then you use to, you doing a scraping attack. Or, you know, if you fly an airplane every day over a factory that’s being built, so you can take photographs from the air, which is actually one of the very early trade secret cases, it feels like you’re acting in a bad manner. Like you’re doing something that isn’t somehow fair.
And a lot of that goes back to how well protected was the trade secret also. Like if you had to go through sneaky, kind of feels icky approaches to ascertain the trade secret, the courts will say that wasn’t readily ascertainable because you had to act in a bad manner to get it. Whereas just doing a quick search or a quick AI query,
Tim (23:26.08)
Yeah.
Chris (23:41.39)
anyone can do it without being dishonest. So there’s just kind of those two prongs of ease of figuring out, like reverse engineering is fine in trade secrets and people do it all the time because they buy the product at the store and then using a screwdriver and a hammer, take it apart and then figure out what’s inside. That doesn’t feel emotionally wrong. Whereas
If I break into your factory at three in the morning dressed like a ninja and steal your laptop and then go home and decrypt it, that feels very wrong. That’s not behavior you want to encourage. Yeah. I mean, you’re breaking lots of laws. Even with AI, you might be breaking the terms of use or the label license that you clicked to agree.
Tim (24:21.899)
Yeah, think that’s breaking the wall in and of itself probably, yeah.
Tim (24:36.555)
that’s interesting.
Chris (24:37.836)
without reading. And then if you, you that doesn’t grant you permission to reverse engineer anything. So that’s, again, that’s acting in a way that is, you know, somehow illegal, breaks contract, breaks the law. You know, those are behaviors we don’t want to encourage.
Tim (24:59.837)
Interesting. so let’s say we’re not, let’s say we’re in a world where we don’t have to, you know, do it in an icky way. We can just go on our computer and ask the question and get the answer. Let’s imagine that day is coming. so talk a little bit about like the time, the time of invention and how that impacts this. so whether it’s, you know, whether it’s.
some sort of obviousness sort of argument or whether it’s readily ascertainable. Doesn’t the standard apply on sort of the date of the invention or something like that?
Chris (25:41.228)
Yes, in the patent world, everything keys off of your filing date. So when you actually electronically delivered your patent application to the patent office is it’s that one magic date that everyone’s crystal clear on. And you’d have to look at what prior art exists on that date. What is the level of skill of that?
Tim (25:47.947)
Okay.
Chris (26:10.574)
skilled artisan as of that date, like everything is keyed off of that one date.
Chris (26:17.912)
But in your world where all this stuff is super easy, just type a few clicks on your keyboard and off you go, it makes getting a patent harder. But it also cuts both ways. What is readily ascertainable in a trade secret world has also gone way up. that’s why early in the conversation I said it’s…
Tim (26:18.037)
Same.
Chris (26:44.652)
Yeah, AI is the rising tide lifting all ships. And it makes it easier for everyone because you have this awesome new tool, but it also makes it harder for everyone because things like enablement and readily ascertainable have risen as well.
Tim (27:00.159)
Yeah. So the idea that I convinced myself of this is sometime in the last year was that this readily ascertainable thing was going to happen. Right. So that’s my belief is that all of this, whether it’s obviousness or readily ascertainable, all of it gets harder for sure. And so the
The way in which you…
think about your trade secrets. So let’s sort of put patents to rest for now. The way you think about your trade secrets in a low barrier, readily ascertainable world order is a much more system level approach. So in other words, if I think about a discrete thing, discrete things are probably going to be relatively easy to predict.
Okay. Whereas the combination of discrete things might be much sort of harder to predict. And so.
Chris (28:12.76)
Yeah, I think that’s right. You have to be more complicated or more of an integrated system approach to not be readily ascertainable. I think that’s right.
Tim (28:24.715)
like system level thinking, right? So it’s almost like you have to kind of take yourself out of this mindset of like, that’s the trade secret right there, right? And you have to sort of say, like, here’s the, it’s almost like it’s biology, right? Like, you know, one cell, congratulations, or one chemical process, congratulations, but do you know how this thing affects this thing, like seven steps downstream or upstream, right? And, you know, again, without having the benefit of
Chris (28:27.064)
Mm-hmm.
Tim (28:54.219)
like lots of studying, lots of knowledge, that’s going to be probably hard to predict, you know, in systems, especially when one of the interesting things to think about there is you also have, have like the intersection of like business data and technical data, right? Of all different sorts, you know, and sort of how they mash together to kind of say, this is the right strategy or this is the right answer. I think there’s something really, really neat there.
Chris (29:21.57)
Yeah, I think so. And you do want to still get the specifics right. And it’s kind of like in patents where we do broad, medium and narrow trade secrets are kind of the same way where you want the discrete trade secrets documented, but then also the larger systems documented. So, you know, all of your trade secrets, whether they’re individual discrete ones or more integrated systems, they all are still.
Tim (29:28.468)
Yes.
Tim (29:37.439)
Yep. Yep.
Tim (29:42.783)
Yes.
Chris (29:51.126)
interesting trade secrets and should be documented as part of your normal trade secret management program. But the discrete ones might be more easily attacked as being readily ascertainable. But that’s not to say don’t document them and don’t count them as trade secrets because they still are and they might have tremendous value.
Tim (29:58.069)
That’s right.
Tim (30:07.113)
Yeah.
Tim (30:14.825)
Yeah, and sorry I’m off on this tangent and sorry if I seem to be talking out of both sides of my mouth. Totally agree. And there’s cases every week that talk about particularity and well, if you can’t define the trade secret, good luck. So that’s very much in play today. That is the world we live in today. So don’t take my system level thinking as anything other than I would start to
Chris (30:32.75)
Exactly.
Tim (30:44.085)
think of trade secrets at both a discrete level and at a system level. But yeah, certainly you’ve got to get the stuff documented at a discrete level based on today’s court cases.
Chris (31:02.498)
Yeah, and think adopting that broad, medium, narrow approach that we’re used to using for patents into a trade secret mindset will help a lot and just go from individual discrete trade secrets and then go broader or go more integrated into a system level will serve you well.
Tim (31:11.881)
Yeah.
Tim (31:23.497)
Yeah. Yeah, it’s actually, you know, something we’ve never really sort of talked about this, but much like in, in the patent world where you have, you know, composition of matter, you’ve got design, you’ve got, process methods, whatever. yeah, we, that’d be neat to sort of like, try to figure out also framework like that. And then, and maybe it is the same framework, for trade secrets, but,
Chris (31:45.07)
here. Trade secrets in my mind, there’s kind of a very clean two buckets, or at least that’s how I think about trade secrets in terms of technology trade secrets, where you know there’s this whole intersection with patents.
Tim (31:53.896)
That’s kind of it. You just sort of spurred that thought.
Tim (32:52.107)
That’s super thoughtful and bears out in the litigation frankly, right? As you just see more business sort of data kind of litigation these days.
Tim (33:20.907)
Yeah.
Tim (33:28.159)
Sure. Yeah.
Tim (33:49.675)
Exactly. Another customer list case.
Yeah. So, one last quick thought, I think going into this discussion, I was really, and you and I are doing some writing in the background on this topic, but this idea that, you know, AI is eating your IP. I think today’s discussion has convinced me that that’s probably true. it’s, and it’s not completely clear if it’s eating your IP.
you know, slowly for now, but it’s going to accelerate very fast or if it’s just, you know, a bit of a, you know, if, if it’s sort of slow and steady, but as you say, you know, this sort of also may, may increase the amount of IP, right? And so, anyway, we may need to, we may need to think about that AI is eating your IP.
Tim (35:00.267)
Okay.
Tim (35:53.363)
Yeah, well said. think we end on that note. And yeah, thanks, Chris. Great stuff. See you next time. See ya.
Find more episodes:
👉 https://www.tangibly.com/insights/podcast
📺 Catch more episodes on YouTube:
👉 https://www.youtube.com/@Tangibly/podcasts
🎙️ Follow us on your favourite podcast platform:
Spotify: https://creators.spotify.com/pod/profile/tangibly/
Apple Podcasts: https://podcasts.apple.com/us/podcast/reasonable-measures/id1745186004
➡️ Subscribe for weekly conversations at the intersection of technology, innovation, and IP law.
#ArtificialIntelligence #PatentLaw #TradeSecrets #Inventorship #FOSITA #Obviousness #IPProtection #Innovation #LegalTech #Tangibly