.png)
KP Unpacked
KP Unpacked explores the biggest ideas in AEC, AI, and innovation—unpacking the trends, technology, discussions, and strategies shaping the built environment and beyond.
KP Unpacked
Built by AI, Owned by You? Maybe.
AI is building faster than the law can keep up. In this episode of KP Unpacked — the number one podcast in AEC — Jeff Echols and Frank Lazaro dive into a messy question every firm needs to answer: Who owns the work when AI creates it?
From copyright law updates to court cases nobody’s talking about yet, they break down where the real risks are for architects, contractors, owners, and innovators. You'll learn why "lazy AI" is a legal nightmare, where the real landmines are for your firm, and why your next big project needs strong AI policies before the first line is drawn.
Key takeaways:
- Why the U.S. Copyright Office says human authorship is still king
- How firms can protect themselves when using AI-generated content
- Where AEC companies are most vulnerable — and why graphics pose bigger risks than text
- Why starting with your own work matters more than ever
- How upcoming lawsuits could change the rules again — fast
Whether you’re drafting, rendering, or pitching, understanding AI ownership could save your firm a lot more than pride.
📍 Two Gen X guys talking AI, risk, and the future of your work — tune in and protect what you build.
🎉 Special Offer for KP Unpacked Listeners: Get 55% off your ticket to the 9th Annual AEC Summit on October 29th at the Diverge Innovation Center in Phoenix! Click the link below and use promo code UNPACKED55 at checkout.
Don't miss this opportunity to connect with top minds in AEC and beyond. Tickets are limited—act fast!
Welcome back to KP Unpacked. This is where the biggest ideas in AEC and AI and innovation they all collide in one podcast. It's powered by KP ReadyCo. This is where we break down the trends, the technology, the discussions and the strategies that are shaping the built environment and beyond, and this is the AI and AEC version. You may have guessed that because, as you're looking at the screen, assuming that you're watching this on YouTube, but if you're listening to it, I'm just going to describe it To my what is it to my right?
Speaker 1:To my right is Frank Lazzaro. He is my teammate here at KP ReadyCo. He is our in-house AI expert. He is not KP Ready. The traditional form of KP Unpacked is KP and I unpacking one of his LinkedIn posts, but this is the AI and AEC version, where Frank and I take one AI topic per episode, we break it down and we give you actionable takeaways for that particular topic. This is we have a lot of fun here and I think we make it very practical. I think we'd make it very useful. At least, that's the feedback we've been getting, and I really enjoyed this conversation. So, frank, thanks for joining me again today. Yeah.
Speaker 2:I'm glad to be back. Yeah, we're getting a lot of good feedback. I think it's timely. I also think too and I think you hit the nail on the head on that one it's, you know, diving deep on that one topic really kind of allows people to kind of really kind of figure out. You know all the things that we hear from not only our client conversations and things that we hear at conferences et cetera, but these are topics that really come from the audience, so it's just as great.
Speaker 1:Yeah, yeah, and that's that's a good reminder, right, as you're listening to this or watching this. However you consume this. It's probably easier if you're watching it to go over to YouTube and leave a comment there on YouTube, but wherever you're consuming this and maybe you're.
Speaker 1:Maybe you found it through LinkedIn. However, you found this episode. If you've got a question, if you've got a comment, if you've got a suggestion for a future topic, just put it in the comments. Our production team not only are they going to provide the links for the things that Frank and I talk about so don't don't panic when we start dropping this, that or the other. You know, look at this and look at that. The links will be in the show notes below, so our production team will put those there but also they'll be looking for your questions, your comments, your suggestions. So if you've got a suggestion about a future episode, hey, we need to know about this. We're just trying this. We saw this tool came out. Whatever it is, put that in the comments below and maybe we'll turn that into an episode in the future.
Speaker 2:That's a great way for us to map out what it is that we talk about here, absolutely. If not, I've got to get to some writing and kind of start writing up some new episodes.
Speaker 1:Yes, please do the work for us. Don't make us guess That'll be my weekend if we don't get any comments.
Speaker 1:Yeah, frank and his wife would like for you to give him a break this weekend. So please leave a comment on what you want to hear, what you want to hear us talk about next week. In all seriousness and that was actually serious, but in all seriousness today, frank, we're going to talk about AI and copyright, which has been a hot topic since, basically since ChatGPT came out, I think, or became available to the public. So we know that AI can generate content could be written, could be images. There's there. It could be code. There's lots of things that we could use AI to generate, and I think it is again my. You know my background. For those of you that may or may not know this, my background is architecture, and one of the things that is on every title block of every drawing set that's in circulation across the aec industry is a copyright statement who owns the copyright on these, on these construction documents.
Speaker 1:So that's one thing right. It's a big deal in the industry. So, when it comes to ai, we're probably not using AI to literally generate all of our construction docs today, but for whatever it is that we're putting together, we need to understand who owns or at least what the copyright issues are and how firms can protect their work. So where do we start with this, Frank? Where do we start with this exploration today.
Speaker 2:So I think we need to start. Let's go back to the beginning. When I say the beginning, though, that's only two years ago, right? So we go back about two years ago when all of these generative AI tools started coming out, right? There was really, you know, probably two or three questions that always came up, and I think we've already covered them right. Like we talked about data privacy, data security last week, we've talked about, you know, using, you know, information and projects, all that other stuff.
Speaker 2:Copyright is one of those things that did come up, and so for the longest time, we didn't hear anything from the government. There was really nothing from the copyright office, and no one's really kind of telling anyone who owns it who doesn't own it. Good news is fast forward to 2025,. This past January, the copyright office actually released a paper that gave guidelines on what is copyrightable and what is not copyrightable when it comes to AI, and so, interestingly enough, they're treating AI just like any software tool. So Autodesk could not claim copyright to a design you do as an architect just because you use their software. Same thing with Microsoft Word can't claim copyright to the book that you wrote just because you use Microsoft Word to write your book.
Speaker 2:The key distinction in all of this is is that there needs to be some human authorship. So if you are significantly writing your own prompts, right, you're not getting the AI to write the prompts. It's not. There's there's no auto creation, so to speak. If you're writing the prompt and it produces something according to the copyright office, you do have some kind of claim over that Interesting right. I think there are varying opinions about that. I think people are like well, is it really yours? Is it not really yours? The government's position is like hey, listen, the existing law is sufficient to cover this. It essentially says that, hey, you can copyright it if a human actually authored it, regardless of the tool that they used.
Speaker 1:Okay, actually authored it, regardless of the tool that they used. Okay, and this is being you know. For those who are, who are wondering about this or question this, this is being tested in the courts right now but, by the way.
Speaker 2:So the testing in the courts, though, is not so much of people coming after creators. They're going after the models for stealing other people's work to train their models. So it's already. It's the existing content owners that are actually suing the models, not necessarily the end users. Okay, so I think that's a clear distinction too. It's like I don't know of any case to where someone is being sued that created something on the backend. Now, I do know, like, recently, openai had that whole Sora thing. There was a Japanese animation artist where, basically, they were basically replicating his work. He sent the cease and desist. They kind of shut the feature off.
Speaker 2:So I think that there's a lot of gray area still. I can't go in and say, hey, create this artist work to look like mine and and then play it off as my own. I think that starts getting a little sketchy. But I think, if you use it for true business purposes, like where you're taking something that you've already written and you're rewriting it or augmenting it or supplementing it or something along those lines, what I think is is that that is copyrightable. There are still a lot of great questions, right, so I have, you know, one big thing I like to put out to you and to the audience is like where do you stand on, jeff, if you just went to chat GPT and kept prompting it over and over again and then basically created a book and then published it, do you own the copyright on that? It's a really good question. I think in some ways you might, in other ways you don't.
Speaker 2:So for those that, um, that do self publish, interestingly enough, if you self-publish through Amazon, when you go to put your book up, there are now, I think, two or three questions that they ask you about your content. They ask you one is any of this content AI generated to to the level to which it is? Is it wholly generated? Is it somewhat generated but edited or only edited? So they so now they're even trying to bifurcate how they're how, how ai is being used. Because you think about it, if I wrote my own book but then also use grammarly to kind of help me fix my grammar which is an ai tool, is that ai so nice? They see, now you start seeing where it gets a little bit confusing. But the copyright office says hey, listen, if you've, if you, if a human has been involved significantly, you can copyright it yeah, well, and, and you know your example of of publishing or or writing, take, take even the notion of publishing out.
Speaker 1:You don't even have to go that far. Well, I guess you do for the sake of argument. But but how many? How many people? I mean, every, every time I open instagram or or whatever, there's always a course or somebody talking about using, using ai to to write a book. You know, hey, become a thought leader, use ai to to write your book, etc. I mean there, I mean there's Frank Kern, who's a famous copywriter, which is a different copyright than what we're talking about, spelled differently. In fact, he's got a course right now and a tool right that he's selling to use that. So I would wonder right this?
Speaker 2:this has massive implications, obviously yeah, I would wonder how many books on amazon are written that way yeah, so it's interesting, so the quote from the report.
Speaker 2:So, by the way, we'll provide you a link to the report so that you can go and kind of read it yourself.
Speaker 2:But one thing that stood out when I was kind of doing an analysis of that report from the US Copyright Offices and this is a quote AI tools used as part of the creative process do not affect copyright eligibility, but AI generated works without meaningful this is where it gets a little gray. I don't understand where meaningful comes in, but meaningful human input are not protected. So what that's saying is is that if and the way I interpret it is that if you don't have a human touch throughout the whole process, it's not copyrightable. But if you have AI assist you in that process whether it's through proofing, whether it's augmenting your existing content or something then it, then there's a protection. That's there. So it is interesting on how they're kind of phrasing some of those words. But yeah, I, I would recommend people you know definitely go check out, read the report if you're, if you're interested in diving deep on this. But you do have some protection even if you use AI to create something.
Speaker 1:Yeah, yeah, so we know, right, it almost seems like they're they're trying to provide job security for some judges, but we know that these things are going to be tested, they're going to be challenged, et cetera, so, but, but we, but we have a feel for who owns AI generated work. Let's just call it that text, images, the proposals. You know your responses to the rfp, etc. What happens, though, if ai created content violates copyright? Do we know that yet?
Speaker 2:no, I don't think we do, and I think that's what's going to play out in the courts right at some point. Yeah, you know, you go back. I think it's going to be less on the word and probably more on the graphic side, because if you're replicating a style or an artist style, I think you're more likely to violate copyright there than it is to holistically lift words, right. So I think you know, and there's ways to kind of mitigate the words where you know, if you use someone's copyrighted work in writing your book, but you provide references to where you pull that information from and you're and you're giving attribution and credit to it Right, that's, that's literally every academic research paper on the planet Right off of someone else's work.
Speaker 2:So as long as you're providing references and attribution, particularly in text written like books, copies, reports, white papers, I think you're going to run less into the issue with that. I think where the struggle is is that you as an architect me as an architect I use AI to replicate some kind of design element that you use, that you've created, that's unique to you. You know some type of fascia or something on the outside of the building, right? It's going to be hard for me to sit there and say that I came up with that on my own right. You start thinking about it. So I think that what you're going to find is, on the graphic side, that this is going to be a bigger issue than on the written word side.
Speaker 1:Yeah, yeah. So, as firms are using AI and again this could be graphic, it could be written, I think it's. In a way, it, the AI, is learning where the data is coming from. How do they ensure that it is original and that they're not in danger of violating some copyright?
Speaker 2:Yeah, so one of the things that I teach in all of my workshops, particularly when we're talking about content creation, the safest place to be is to start with something that you've already created Right. The safest place to be is to start with something that you've already created Right. Our, our industry is historically and almost comically known for this concept of go buys. We never create a proposal from scratch. It's always let's go back to that previous proposal and then we can improve it. Let's go back to this previous design and we can improve. It's almost kind of like we love the concept of go buys. Well, that creates a lot of efficiencies in most of our processes, but it's In general.
Speaker 2:If you're looking to avoid any issues, if you start with your own content and then look to augment and improve and all this other stuff, what you're going to find is that you're less likely to run into those kinds of issues.
Speaker 2:And that's even what the Copyright Office is saying is that if you have significant human input into what you're creating meaning that I'm using AI to help write a book into what you're creating, meaning that I'm using AI to help write a book I'm going to write the most of the chapter myself and then use AI to clean it up and edit it and really kind of make it flow a lot better. That's copyrightable, according to the copyright office, because the work really generated for me the genesis of the work is for me. Ai is really just helping me kind of clean up the flow and editing. So if we look at it from that perspective, you'll avoid issues if you start with stuff that you already created, versus saying, from whole cloth, create me something. And I think that's where you're going to run into issues. So if you're being really super lazy and you're just having AI do all the work, then you're going to really kind of run into issues, in my opinion.
Speaker 1:Yeah, yeah, I mean that's. I think that's an important point. Right, we can. I think we talked about we probably talked about this on um, on a lot of our episodes right, you have to apply critical thinking, um, lazy.
Speaker 2:AI is bad, yeah, and I think it also plays into the point of that. It reinforces the fact that AI is not going to take your job because there has to be human involvement at some point, when it comes at least to the generative AI side. You know, on the robotic side, you know there doesn't need to be human involvement. You basically program it and it uses AI to kind of do a task, but you're not copywriting those tasks. What you're really copywriting is the creation part. So if you're thinking about just creation in and of itself, start with your content as a base and what you're going to do is you're going to find that it's going to be less and less of an issue and you're not going to. Now, again, I think from a written word perspective, I think you're less likely to run into that issue.
Speaker 2:Where does this go? When you start thinking about video, you start thinking about rendering, you start thinking about photos, to where, essentially, all of those models are built off. Basically, they're training it off of other images. So those images are coming from someone or somewhere. They're not necessarily all open source or somewhere they're not necessarily all open source. So if I'm going in and wanting to render a building in a certain particular style, right, I, like you, know Frank Lloyd Wright, right? Is there a chance that I could be infringing on some kind of copyright or design element or something along those lines? If that's what I'm asking him to do, like I'm asking the AI to do that, like it's not going to randomly just go find some minimalistic design that's not his, yeah, right, it's going so. So, again, I think, images and video. I think we're going to run into more issues, mainly because of of where they're finding their content to train the models yeah, yeah, there there's that famous video from um.
Speaker 1:Yeah, there's that famous video from ZHA, which is Zaha Hadid Architects, from their computational design studio, where they basically the data that they have, the AI, go through all kinds of iterations and gyrations and create building after building, after design, after design after design. Some are horizontal and some are vertical, and so on and so forth. This video that comes out of their like I said, out of their computational design studio. You watch this video of the AI. If they call it hallucinations maybe that's that is what they call it but as you watch this, you see what could be considered, project after project after project.
Speaker 1:But every single one of them, if you know their work, if you know zaha's work, you look at it and you go, yep, zaha, zaha, zaha, zaha zaha. And in fact, none of them are actually quote-unquote, designed by zaha hadid architects. It's just ai creating all of these based on a depth of data you know from, from the style of of their work, from the style of their design. So that's that's one. That's that's uh easily, uh, uh, searchable on on youtube. You can find it pretty easily.
Speaker 2:We'll put it in the show notes yeah, there was another case that was recently decided, I think in february of this year. Um, you, and essentially it's revolving around, you know, I think the federal courts, you know, are really kind of focusing Cause again. This is like the lawsuit someone someone was trying to copyright something that was 100% AI generated and they were denied. So it was one of those things to where, because there was cause, it lacked any human authorship. It's like you program the AI to do something and AI did it, and then he tries to copyright it and it's just like, well, wait a minute, you, there was no human involvement in that creation. Yeah, right, so I think I think it's, I think it's more complicated. I think this is kind of a. I think the way the copyright office has addressed it right now, it's a little bit of a bandaid. I think again agree with you 100.
Speaker 1:I think this will eventually kind of be driven either through policy changes or lawsuits or both yeah, yeah, I guess it's like anything else, um, you know, any sort of new territory it's, it's gonna have to be uh, tested and challenged before, before it's all settled out so.
Speaker 2:So I think the existing content owners are the ones that are really upset right now because, again, these models are only as good as the data that they can train on, and so what they're doing is that I don't think people understand this, but they're essentially just sucking in as much information as they possibly can, and part of it is copyrighted information, right. And so then the question becomes you know like? To what benefit or to what end are they allowed to do that, right? And you know, maybe the content creators may end up losing. At the end of the day, they may end up winning, I don't know. So it is. It is fascinating to kind of see how that plays out.
Speaker 1:Yeah, I agree. And so as we think about this, you know, we head towards the part of our episode here where we give some takeaways, some actionable things. Do you know, based on the information from the Copyright Office and other cases that you've heard about, what is it that firms need to do? We know that they need to be careful right, we know all those things. We know that there's plenty of gray area around the ruling or the paper, I guess from the copyright office. But what is it that firms need to do to protect themselves? I think on the side of making sure they don't violate something, but also on the side of how do they protect themselves?
Speaker 2:Yeah, yeah. So I think there's a couple of things. One and I think we've covered this before in a previous episode right, Strong AI policy is probably a very good place to start. Right, Just like with everything else, you know the do's and don'ts of what you can and cannot do. The other thing, too, is you know, within that policy, or particularly within the training that you're offering to your employees, because I think training is a very big, important, big again, I think the concept of training is just reinforced by this concept as well.
Speaker 2:Right, Like you're not training people there, they can inevitably do something. Maybe that's an inadvertent way, but they could do something that could run afoul of copyright. But the way I look at this again is go back to, if you're using your own content to create new content, some modification of that. You are in a better position to be protected that if you just asked it to kind of create something willy-nilly, a whole cloth type of perspective. So really, two big takeaways for me is that, one, make sure that you have the right policy in place. Two, you need to train on that policy so that people actually know how to use the tools appropriately. And then the third component is really just focus on making sure that the content that you're using is starting with something that you've already created.
Speaker 1:Yeah, yeah, I think those are good points and, yes, there's going to be a chicken and egg there, right? Well, that's great, frank, but we're just starting out with AI. How do we base it on something we've already created? Right, there's a beginning to this right that you have to, that you have to build from and and I agree, I mean, I think it does so much of this in terms of protecting your firm. I think it does very much, just like you just said, it does go back to having solid policies and trainings. I don't, I don't see how going forward and right so you know many people that listen to this know that I teach professional practice.
Speaker 1:This is going to have to become a component of of pro practice classes too. Right we have. We have to have policies that are designed to allow us to use the tools properly and keep us out of trouble essentially. I mean, that's way boiled down?
Speaker 2:I think yeah. And what I do think, though, too, is it becomes more and more important as, like right now, a lot of these tools are third party tools, that kind of sit. You know, you have to go to chat, gpt, you have to go here. How does that change when it's actually when the ai is actually inside autocad? How does that change? When it's inside revit, how does that change? So you have all these design tools now that either are dabbling in the AI and they're doing something. But let's fast forward four years from now, five years from now, when AI is truly embedded in those applications. How do you protect yourself? Right, because right now it just kind of sits on the outside? Right, I can go here, I can go there, but once it gets into our everyday apps as an architect or design professional or something along those lines, I think that's where, like you said, you're going to have to integrate it into your curriculum. But I think that is something that most firms are going to have to deal with, that going forward.
Speaker 1:Yeah, that's. That's an interesting point because you know when. One of the one of the questions that comes up a lot is well, when is Revit going to Revit whatever 236? When is it going to be rolled out with full AI capabilities? And there's all kinds of reasons for Revit being where it is today. But I think that point that you just made is an important one when is the data going to come from?
Speaker 1:Because, if you use a Revit as a, for instance I'm not picking on Autodesk If you use a building information modeling tool that has full AI capability, where is the data going to come from? Is it going to come from your work, from the work of your firm, or is it going to come from a larger data lake that's housed at the mothership, at the mothership? Because that, that also, that this, this opens up a whole nother episode, because the liability, the potential liability there is huge, right.
Speaker 2:Well, there's also the question of like yeah, the other thing is you start thinking about the concept of synthetic data. Well, yeah, so you can. You can have an AI kind of create synthetic data from you for you, which is really not owned by anybody. But then the bigger question is just like where did it learn to create the synthetic? So it has to learn it from somewhere.
Speaker 2:So at some point there has to be some source, there has to be something at some point that's telling it what those elements are, so that it could create synthetic data. So I think it is a bigger question. I think that one. I don't really believe that there is a true answer right now. I think we have, we have a guiding light on what is acceptable and what is not acceptable in this, in this area, but I still think that there's more to come when it comes to this particular topic.
Speaker 1:Yeah, yeah, absolutely, and, and so we'll, we'll. We'll wrap this episode here in our conversation on AI and copyright, but I'm sure that we will revisit this again very and then it also starts to morph into other conversations about different types of liability as well, so we will definitely be back to this topic and tangents from this topic in future episodes. So, again, we appreciate you listening to this. Let us know what you think. In the show notes, um, in the the comment section, especially if you're watching this on youtube, give us questions, give us comments, give us recommendations for future topics and, uh, our, our production team will be looking for those, and also our production production team will be dropping links for things that we're talking about into the show notes. So look for those. As always, my name is Jeff Eccles. I head up our mastermind program and our incubator here at KP ReadyCo, and I'm joined by my teammate, frank Lazzaro, who is our AI guru. Now, he's not an AI-created guru, he's our guy that's staying on top of AI. How's that? That's right.
Speaker 2:I'm the Gen Xer that just happened to figure out AI.
Speaker 1:There you go Two Gen X guys talking about AI. That's the new name of this show. So, frank, thank you, as always, for joining me for this episode today and thank all of you for listening. We appreciate youed the AI and AEC version. This is where the biggest ideas in AEC and AI and innovation collide. It's powered by KP ReadyCo and this is where we talk about the trends, the technology, the discussions and the strategies that are shaping the places where you live, where you work, where you play, where you worship, where you eat, where you do all the things that you like to do the built environment. So thanks for joining us. We'll see you again next week. Thanks, everybody.