The AI Smoking Gun: Will a "Non-Inventive" Report Sink You in Litigation? w/ Robert Plotkin

00:00:06: Hey, what's up?

00:00:07: Welcome to Beyond Best Practice.

00:00:09: I'm your host, Bastian Best.

00:00:12: What happens if your AI search tool concludes an invention is non-inventive?

00:00:17: This question came up recently in my seminar where I teach other patent attorneys how to draft patents with AI.

00:00:24: And it's the smoking gun scenario, right?

00:00:27: So you make a patentability assessment with AI support.

00:00:31: The AI says non-inventive.

00:00:33: You disagree with the AI.

00:00:35: You get the patent.

00:00:36: And then years later, the AI report is found in discovery in the US litigation.

00:00:42: So what happens?

00:00:43: To get a definitive answer, I took this to Robert Plotkin.

00:00:46: Robert is one of my go-to US patent attorneys for AI patents and an authority on the subject.

00:00:53: And here's his take.

00:00:55: This is such a great and relevant question, and I'm really glad we're talking about it because AI tools are becoming integral to how so many of us practice patent law right now.

00:01:08: But here's the thing, although the technology is quite new, the underlying legal principles and the risk management strategies that you would apply to them are actually quite familiar to us as patent practitioners.

00:01:24: So let me ground this in what we already know and what we already do in analogous contexts.

00:01:32: First of all, obviousness and patentability generally are always determined claim by claim based on the actual specific wording.

00:01:45: of the claims.

00:01:46: So that's the first thing to keep in mind.

00:01:48: Second, we already navigate how to deal with preliminary patentability assessments in the form of search reports, our own internal analyses, and so forth all the time, even before and outside the context of AI.

00:02:09: And as a result, Patent practitioners and patent offices and courts already know how to weigh and take into account different types of preliminary patent ability assessments, even if they are later overcome or practitioners changes their minds or take different actions later on.

00:02:33: during prosecution.

00:02:34: So when we think about this scenario, it's really not that different from situations that we already handle every day.

00:02:43: You know, think about when you get an initial search report or office action that rejects claims based on lack of novelty or lack of inventive step, obviousness.

00:02:57: And then you, as a patent practitioner, overcome that rejection, you amend the claims, or maybe even you don't amend the claims, you distinguish over the prior art.

00:03:08: Or consider preliminary patentability opinions that we generate internally or receive from searchers that we hire, and then our own opinions evolve as we refine claims during drafting of the application or during prosecution.

00:03:30: You get international search reports or initial search reports from patent offices and any of these types of reports might flag references as highly relevant or novelty destroying and then you later distinguish over them.

00:03:48: Internally before you file the application or during prosecution.

00:03:52: right this happens all the time.

00:03:54: so the key point I think that really matters here is that.

00:03:59: One, we already deal with this, and it's not necessarily a problem.

00:04:03: And if it comes up later in litigation, you have many ways to argue that your own decision to continue prosecuting wasn't contradicted by a search report or some other kind of initial assessment of non-patentability.

00:04:20: And we also know that the invention is something that's always defined on a case-by-case basis based on the language of the claims.

00:04:31: It's the specific claim language that matters.

00:04:33: So in going back to the specific question that you asked about an AI-generated assessment, which concludes, so to speak, that an invention is not patentable.

00:04:46: Let's say you generate that or you receive such an AI-generated report.

00:04:53: that's based on some general description of the invention and the AI says that the concept or the invention is not inventive.

00:05:02: Well, that may or may not be relevant to the claim that is later in prosecution or in litigation because the claim has its own specific language.

00:05:14: That in almost all cases, you'll be able to distinguish over whatever the AI said.

00:05:25: And as I said earlier though, even if the language in the claim that you're prosecuting or that's in litigation is exactly the same as what the AI said was non inventive, I don't necessarily see that as a problem.

00:05:38: It's not really different from.

00:05:41: All these other cases I mentioned that we deal with where you get a search report from a searcher or a patent office or an office action and you're able to distinguish over that later and argue in litigation that you had your own reasons for continuing to prosecute despite that.

00:06:00: So, but in addition, there are some risk mitigation steps you can take that build on practices that I think many of us already use.

00:06:09: so you can document.

00:06:11: why you are arguing against or traversing rejections from an examiner.

00:06:16: You can document the basis for a specific claim language that you write.

00:06:23: You can document the reason why you wrote claims in a certain way in light of an internal patentability analysis.

00:06:32: And of course we all know there's pros and cons to such documenting.

00:06:36: By creating that documentation, that documentation might end up being discoverable in litigation later.

00:06:42: So you have to think carefully about the pros and cons.

00:06:46: That documentation might help you, it might hurt you later.

00:06:49: I know a lot of practitioners who decide not to ever create such documentation or to routinely destroy that.

00:06:59: documentation, let's say at the time of filing the patent application, I think there's good arguments to be made in both cases.

00:07:05: Most litigators would tell you, I think they, who are representing you would want you to not have such documentation, but I just mentioned it as a possibility.

00:07:18: But the key thing is, as I said, Claims are going to evolve over time though language is going to evolve over time and I think in almost all cases You you would be able to point to How specific claim language that's at issue in litigation differs from Whatever the AI said was non inventive and that's one basis for defending your position and litigation.

00:07:54: But I'm just gonna say it again, even if that's not the case and the claim that is granted was the exact claim that an AI system, quote, concluded was not inventive.

00:08:08: In the end, the judgment about what claims to file, what claims to argue for, is yours as the practitioner.

00:08:23: You exercise your own judgment, and I think if a claim were ever litigated, even if it was exactly identical to something that an AI system had said was not patentable and you were deposed or had to testify in litigation, you could explain why you brought your own judgment to bear on that claim and decided that it was valid to pursue such a claim in prosecution.

00:08:53: And again, that's not any different from what happens very often when we get search reports internally or do preliminary assessments of patentability and move ahead.

00:09:07: in a way that might seem contrary to those preliminary assessments and often how we prosecute in the face of international search reports or other kinds of search reports from patent offices.

00:09:21: We bring our own professional judgment to bear and moving forward and that is.

00:09:27: as long as you're being thoughtful about it, that's something that you should feel comfortable justifying later in litigation if it ever comes to that.

00:09:37: So I think that the bottom line here is that although AI brings some new capabilities to our practice, which can be very valuable, the fundamental framework for handling preliminary patentability assessments of various kinds and for managing privilege and for defending patentability conclusions that you draw, all remain remarkably consistent with how we've already handled these issues long before AI came about.

00:10:10: That is such a practical and reassuring answer.

00:10:14: Basically, what I take from it is this.

00:10:17: This is just a new AI powered wrapper on a very old and familiar problem.

00:10:23: We already handle things like preliminary search reports and office actions that we disagree with and an AI generated opinion is in principle no different.

00:10:35: If you want to connect with Robert Plotkin you can find the link to his linked in profile in the show notes.

00:10:41: And speaking of AI in patent work, Robert's advice is all about integrating new tools without breaking old established legal principles.

00:10:50: If you're looking for tools to help you do just that, check out the free AI tools for patent attorneys on my website, powerclaim.io.

00:10:59: You can go there to try them out and see how to supercharge your own practice without creating new risks.

00:11:05: That's powerclaim.io.

00:11:08: That's all for this week.

00:11:10: I'm your host, Bastian West.

00:11:11: Talk to you next time.