The Hidden Copyright Risk in AI-Generated Content—and How to Protect Your Business with Annabel Kaye, Founder of KoffeeKlatch | Ep 234

Dr Steve Day: When you use AI to repurpose, create or improve content, who owns the copyright for the work that\\\'s produced? Is it you, is it the AI, is it the virtual assistant you got to use the AI or something in the middle? That is the topic of today\\\'s episode, and I\\\'m joined by Annabel Kaye, who\\\'s the founder of KoffeeKlatch. Who helps small businesses with legal contracts and agreements. Such as data protection, copyright and virtual assistant hiring agreements.

This is an area which was totally great to me and I was quite surprised at some of the stuff Annabel told me on the episode. And do stick around to the end because we\\\'re sharing a link. So you can get some discounts on Annabel\\\'s services. And I also share some systemised approaches to actually implementing the stuff she shares in this episode. And remember to hit subscribe so you don\\\'t miss out on future episodes on all things cool. To help you create a freedom-driven business that gives you the lifestyle you deserve.

Annabel, thank you so much for coming back on the show. I\\\'m excited to chat with you today after receiving an email from you, actually. We were just talking about this before how not very many people who emails actually open these days and mine all filtered by my assistant. But that yours is one of the ones I get her to forward because I always find it interesting. And sometimes very relevant, and this was one of those cases we\\\'re talking about.

It was talking about AI, about copyright, about the gray areas of this at the moment. And I thought what a great opportunity to get you back on the show and share your wisdom again. And hopefully cast some light onto this rather sketchy area of copyright and law at the moment. So hi, and welcome.

Annabel Kaye: Thank you for having me back. We always get really technical things by the way, don\\\'t we? And yeah, well. The problem with AI and copyright for most entrepreneurs is most entrepreneurs have got very shaky grasp of copyright when it doesn\\\'t involve AI. So if you like, when you\\\'ve got a change or a potential change that builds on something you didn\\\'t understand. It\\\'s much more difficult to adapt to it than if it builds on something you did understand axiomatically really, isn\\\'t it?

So what most entrepreneurs don\\\'t get about copyright, and this is true in the UK. But it\\\'s true in most of Europe, and I believe in America, it may not be globally true. Is that whilst copyright in your works belongs to you, if you pay employees to create stuff, that copyright still belongs to you. If you pay an outsourced VA or graphic designer or web designer to do stuff. The default on the copyright is it belongs to them unless it is assigned to you. Which can be done through contract. It can be done in Germany only through specific types of paperwork. There are various ways it can be done, but we tend to think as entrepreneurs because we paid for it, we own it.

But actually what we\\\'ve paid for is the right to use it. And the right to use it is not the right to repurpose it, edit it, sell it on all these other things. So because we\\\'re in a muddle about that, we are completely rear ended. When it comes to understanding the effect that AI has on copyright. Because if something is 100% generated by AI, by a bot or an app that you\\\'re using, there have been two schools of thought on who that copyright belongs to. Because obviously AI is not your employee and it\\\'s not you. So you can see where I\\\'m going with this.

One possibility is that copyright belongs to the computer code programmer who created the bot. Because a bot is not a legal person. Is it? Another possibility, which is in the early days of AI was muted, was that there was no copyright. That because a bot is not an entity short of giving data in Star Trek type rights to AI bots. It can\\\'t hold copyright because it doesn\\\'t exist.

So for the last couple of years that\\\'s been a bit of a debate. While we\\\'ve awaited draft legislation from Europe, America, all the rest of it. It\\\'s never quit. And at the moment the balance of view which may change is that bots don\\\'t exist. They don\\\'t hold copyright. So what this means is if you are generating stuff 100 on bots. And it\\\'s part of your brand, it\\\'s part of your IP, it\\\'s part of your content. And someone else copies it, you can\\\'t say you can\\\'t use that stuff because you didn\\\'t own it.

Dr Steve Day: Can I ask a question? This may seem obvious. But how would anybody know whether it was you who created it? Or it was a AI bot that you used and then created it? Because the AI bot\\\'s not got, not going to come chasing you. Well, not yet anyway. The AI bot\\\'s not going to come chasing you. And getting annoyed about infringement of copyright. Which is what obviously is the risk if you copy somebody else is a human\\\'s work. Is that one day they, they notice that you\\\'ve copied them. And they come knocking with their lawyers.

Annabel Kaye: I don\\\'t think that\\\'s the risk. The bot won\\\'t care. Right. But supposing I produce a KoffeeKlatch branded download on what you need to know about transferring data overseas. But I\\\'ve had no human input in that whatsoever, and I put copyright KoffeeKlatch on it because don\\\'t we love to do that stuff? And I send it out in the world and two months later someone is circulating an identical download. And the only difference is they put copyright big rival on it. You know. And I go, how dare you use that material, it\\\'s mine. And they go, prove it, and I can\\\'t.

Dr Steve Day: So it\\\'s in that element of proof then. So it\\\'s like you\\\'d have to sort of have your work...

Annabel Kaye: I can\\\'t get a takedown order for breach copyright. I can\\\'t do anything. And although in some areas of work you\\\'re going to get similar things. You know. If we\\\'re explaining a similar law, we\\\'re going to use a similar structure. But bots can produce the same structure, they can produce all sorts of things. So it\\\'s going to get very murky.

And obviously if I\\\'m in the content creation business and I use a 100% bot-generated content. And I say to you, I\\\'ve done all this work for you, Steve. You owe me whatever. And I\\\'ll assign you the copyright, and I never owned it because it was 100% bot. You know. I\\\'ve committed fraud because I\\\'ve assigned to you something I don\\\'t own. And that\\\'s going to be a problem in people who were trying to outsource things. That help build their IP, help build a scalable salable business.

Because when you get to due diligence, they\\\'ll go, do you own all the copyright? And you\\\'ll go, yeah. And they\\\'ll go, well, show us how you know. So for some people it may never be a problem if a bot creates a social media post for me tomorrow. It\\\'s ephemeral, isn\\\'t it? Yeah, it doesn\\\'t really matter for other parts of what you\\\'re creating in your business, it\\\'s a problem.

Dr Steve Day: So just, just to sort of, for me to weed this. Get through the weeds if you like, for my own peace of mind. So in, in my work, for example, like most of what we produce now. No. I\\\'ll actually say all of the copy I produce now definitely has a helping hand by AI.

Annabel Kaye: Absolutely.

Dr Steve Day: Content that we produce. So the subject matter, the knowledge and experience that\\\'s coming into the what of the content. That for me comes from typically my podcast. So I\\\'ll record podcast and then I\\\'ll get AI to transcribe it. Still my copyright, because just my words. But then I\\\'ll get AI to take my transcript and to apply my style guides, my, you know, everything I put into my GPT agents. To convert that transcript into a post that\\\'s written in my tone and voice.

Annabel Kaye: Right. I think that\\\'s not uncommon. And I think the other common use is around the other way of maybe getting it to do a first draft. And then putting your own spin on it. You know, those are the two biggest juices I\\\'m seeing in my own clients. And there\\\'s nothing wrong with that, by the way. I\\\'m quite pro AI, but in that sense, what we\\\'re talking about is not AI-generated, but AI-assisted.

Dr Steve Day: Yeah, okay.

Annabel Kaye: And if you like, in the area of assisted by AI, the issue is, did the human make, and this is a quote from various. You know. Consultative bodies and laws, a significant input. Right. So if it produced a draft and you just went, I want the spelling to be American instead of English, or vice versa. That\\\'s not really a significant input, is it? So you\\\'re going to have to look at if it\\\'s. If it\\\'s really important, what does that mean? Now, we\\\'ve only had one case determining this, which weirdly enough, is in China. And before you go, what they think in China\\\'s got nothing to do with us because AI is a global thing. Copyright\\\'s becoming a global thing. It gives us a clue.

And that was what I was emailing you about. And it was very much in line with how copyright is determined in the UK anyway, ignoring AI. So let\\\'s suppose you and I had a copyright dispute over something simple like a book, right? And I said, you nicked my book. You just took my chapters and adapted them. And I went to an IP lawyer and you went to an IP lawyer. And what would they ask us to prove copyright.

Well, I know from mine, because I\\\'ve had to do it a couple of times. They go, well, we\\\'d like to see the drafts, we\\\'d like to see how you started it. How you changed it, what process you went through and. And where you got to in the end. Now, as the author of a book, that will be a natural process if the only thing you\\\'ve got is a book. One draft which starts with my copy, or vice versa. And then the amendments is fairly obvious who created the book, isn\\\'t it?

Dr Steve Day: Yeah.

Annabel Kaye: So the answer with a lot of this lies in record keeping. So if what The Chinese case said, which was very consistent with that. Which is why I thought it was interesting. Was if you\\\'re doing stuff that\\\'s intellectually important, keep a record. You\\\'ve got your styles on ChatGPT, so you could go, this is the things I always use. But for important stuff, keep a record of the prompts and keep the drafts. So if it started with your verbatim transcript and you changed it. Or it started with your idea and it wrote it. You should be able to see the iterations. Now, I\\\'m not saying for every post, but for anything important. So what we are doing progressively. We finished the UK round, we\\\'re starting on the global.

It\\\'s been busy. Is requiring people who create stuff for you to keep those records and hand them over if asked. So that if I then go on and publish a blog post and you go to me, Annabel, you stole it from my webpage. I can go, hang on a minute, this is exactly how it was produced.

Dr Steve Day: Yeah.

Annabel Kaye: Now, I say hands it over if asked. Because it shouldn\\\'t be a big deal to keep working records of what you\\\'re doing. I\\\'ve got working records of draft contracts that go back to about 2018, reflexively. Because I need to know, why did I change that? What went on? I don\\\'t keep working records of every social media post I ever made. We\\\'d all be bored to tears.

Dr Steve Day: I do, funnily enough, I arguably have records because I saw everything digitally. I work in Google, I save everything to the Google cloud, I don\'t go through and delete stuff. So I arguably probably have every drafts of every email, every social media post, everything I\\\'ve ever done. Until AI ChatGPT started will all be saved in there. And now, because I got a paid ChatGPT account, never deleted anything from my GPT account. And I use Fathom AI, another AI tool. But so Fathom AI to record all of my meetings. And they apparently will store them indefinitely. And so, you know, that sounds like actually I\\\'m doing easy enough to defend myself should a case arise against me.

Annabel Kaye: Absolutely. And not just to defend yourself, because not just defensive. But if you saw someone doing something. That blatantly copied what you were doing, you might be able to stop them doing it. Because you\\\'d be able to show that you generated it. And in order to. For them to show they generated it, they\\\'d have to have the equivalent record.

Dr Steve Day: Yeah. And the great thing about digital stuff, which I\\\'ve actually sadly had to use. Because I had had to prove that one of my staff wasn\\\'t doing the work they were doing. And one way I could do that is actually all the audit trails within documents that are still digitally. Like for example in Google, you can actually go through. And see every single change that was ever made in a document at any time by anybody.

The IP address of the person that changed at the time. Date stamp, et cetera, et cetera. So if you need to go back and actually prove stuff. If you are drafting stuff, I\\\'m sure if you\\\'ve got notebooks, it\\\'s the same principle, you know. Oh yeah, hard pressed to go and recreate 10 years worth of notes in date order in a notebook. So to prove, to try and cheat the system. So, you know, same thing, just whatever it is you\\\'re keeping that actually do.

Annabel Kaye: Keep it, because if it\\\'s valuable, you don\\\'t. You know. I think the problem is when you think that the person you\\\'re paying is doing the work. And they\\\'re not, if they\\\'re doing 100% AI and you don\\\'t know. So absolutely, I think it\\\'s going to be established that with significant human input you\\\'ve got copyright. And this is the first case on what significant means. But it was very much in line with how we prove copyright in the UK anyway.

And to me it makes sense because you don\\\'t really want another method of doing things. Of course, for the people who don\\\'t do anything. And we get their emails sometimes, don\\\'t we? And we don\\\'t read them right, there is no drafting. It\\\'s like, tell ChatGPT, write me an email about bang and it\\\'s off. Mind you, I don\\\'t think I\\\'d ever want to replicate that. But if ChatGPT replicated it again. Which it might do on some random spin the wheel basis. You couldn\\\'t really complain that someone else sent out an identity email.

Dr Steve Day: No. Can I bring this? I like talking about really practical stuff when we\\\'re talking about things like this. And so one thing that we did. I actually recorded a podcast, podcast number 65 if you\\\'re interested. Which dates back a good couple of years now. But exactly what you talked about at the beginning, Annabel. Which was about the copyright assignment. So I can\\\'t remember how I came across this.

But I realized that actually any work that my virtual assistants are doing for me. If they were based in the UK, it would be black and white. That work would not be owned by me. I think it is gray how that would work across seas that we mentioned before. But even so, I felt, look, this is actually an important thing to at least put some things in place. So I now as a stand apartment onboarding process, just have a simple copyright assignment document. Which I assume you have something like that on your site.

Annabel Kaye: It\\\'s embedded in the VA hiring contracts that we issue. We try not to have them all as separate documents. Because what we with not just the people you send us. But the entrepreneurs we get generally is they\\\'ve got way too much to do to remember to issue document number one. Document number two.

Dr Steve Day: I probably missed that because I think I\\\'ll have to get. I\\\'m now using your document so I can actually probably ditch.

Annabel Kaye: Have a look, have a look.

Dr Steve Day: We\\\'ll go through and when you\\\'re interested. Just while we\\\'re on the subject, if you go head over to sys.academy/tools, you\\\'ll find a link to KoffeeKlatch. Which is Annabel\\\'s company. Where you can get a discount for signing up and getting things like VA contracts. And like we just talk about copyright assignment contracts. And those have recently been updated for the whole AI work. Is that correct, Annabel?

Annabel Kaye: The UK range has their global wages. Next up. We were actually waiting. What we\\\'ve been waiting on, which we now know is coming out in the autumn. Is what\\\'s happening to cross border data transfers. Which I know to most people it\\\'s not an exciting piece of news. But it is for us because what we try and avoid is doing updates on the updates with people. Because when you\\\'ve got all your team contracted, you don\\\'t want to go and the contract of the month is. Yeah, you know. But for people in support. And your minimum support period is six months, you get free updates anyway, so. And longer if you pick different levels, you know. So we now know what we need to do. We already know what we need to do about AI copyright.

We now have a good idea what we need to do about cross border data transfers. And the biggest problem we have. By the way. In terms of user experience with KoffeeKlatch contracts is an interesting one. If you\\\'re on top of how data privacy works in the UK, even if you\\\'re a micropreneur. If you understand what a data privacy policy is. What data sharing with the third party is. The whole sharing it overseas in the Philippines thing is just an add on. You know.

What we find and I\\\'m not knocking anyone for this, I\\\'m sure I\\\'d be the same if it wasn\\\'t. What I do for a living is we get a significant subset of clients who know nothing about that. Who therefore fall into this brick wall of why are you sharing this data? And instead of having one of the currently six statutory reasons go because I want to. So they\\\'re kind of off page, you know, they\\\'re like pre2016 trying to deal with what are going to be 2025 updates. That\\\'s a big gap. So one of the things, and I was talking to someone else about it was. Because we\\\'re entrepreneurs, we like to be last minute dot com, don\\\'t we? You know, we want to go, I\\\'m way too busy, get me a VA, get me a VA now, get me started on Monday.

And then you go, I need to fill in the contract, and then you, I don\\\'t know what I\\\'m doing. And I think it\\\'s worthwhile if you\\\'ve got the systems for it and maybe it\\\'s something you can help people build. To build again. When, are you going to need to outsource module? Because if you leave it to the last minute and you randomly hire someone because you\\\'re busy.

And then you randomly attempt to handle cross border data sharing. Particularly landlords and property people with ID and all sorts of really significant data. You kind of hit a brick wall of I thought this was going to save me time. But actually just right now this is costing me a lot of time. Because the time you should have spent over a year, maybe two years is now having to be spent in a week when you get this sorted out.

Dr Steve Day: Yeah.

Annabel Kaye: And it\\\'s interesting because we have people who go through this. No problem, fine. And I\\\'m doing. And you get other people going, what do you mean? There are rules about what information I can share with whom and they find it a little bit tricky. And I know you\\\'re great on systems. Have you ever built a system for. How do you know when you\\\'re approaching that outsourcing moment. And maybe not get there at one minute to midnight?

Dr Steve Day: We do. It\\\'s called your personal systemization plan. It\\\'s module one of our free time course. There you go.

Annabel Kaye: Well, obviously we need to get that out there to more people. Maybe we should talk about getting that out there. Because we get nothing to do with that course, you know.

Dr Steve Day: Yeah. But what we also we don\\\'t do, which is really good idea explicitly what we don\\\'t do. Is have a step by step getting ready for the data protection side of that as part of that process. We do have access to your, to your site. Now I actually worked with a data protection guy a couple of years ago and he provided some access. But it was generic and it was overwhelming. The stuff I felt like for most people it was just too much too soon.

Whereas I think actually just putting some simple, like, common sense steps in place. So you\\\'re actually moving towards. Towards actually being in a position that actually, when you do need to think about it, it\\\'s actually not a big deal. Like you said, you\\\'re educating yourself towards the goal rather than trying to do it all in one go. That is the key. And a lot of what we teach anyway, the principles. Because I\\\'ve been through this type of stuff myself and the training.

So how we, for example, get people to set up their. The sharing of the shared drives and how they store things digitally. And the sharing of passwords and the documentation and the audit trail. So all that kind of stuff we actually just get people into is because I\\\'ve been through this. And actually then realize, wow, I\\\'m actually really in breach here, or, oh, I\\\'m really unprotected here. And so that\\\'s the way we guide people anyway. But it\\\'s not formally done with a view of, hey, this is also going to help you towards this data protection goal as well.

Annabel Kaye: People don\\\'t realize that they need to look at their data Privacy Policy statement if they\\\'re sharing data. Because VAs, self employed VAs, whatever country they\\\'re in, of third parties, you can\\\'t just let them all in and hope for the best. And if they\\\'re overseas, you need the right paperwork. But people need to know. And I certainly find in the landlord space and the property development space. Where we get a lot of people using VAs in the Philippines. Not necessarily from your customer base, but generally.

Dr Steve Day: Yeah, you know, we have lots of property, yeah.

Annabel Kaye: They\\\'re very, very keen to email. Or scan of someone\\\'s passport to a person at the other end of the planet. And when I go, that\\\'s not secure. Where\\\'s your data processing impact assessment on how dangerous that is? They go, what assessment? Yeah, I shouldn\\\'t have to do this. I\\\'m in a hurry, I\\\'m busy and it\\\'s like, I wouldn\\\'t. I mean, I\\\'m lucky, I\\\'m nobody\\\'s tenant, you know. But if anyone ever wants my ID, even though it\\\'s a legal requirement. My first question is, who are you sharing it with. And how is it being shared and stored?

Dr Steve Day: And also these days, like, there are apps. Like, there\\\'s one I can\\\'t remember my lawyer used last time I had to provide documentation. Seritas I think, or something like that, it\\\'s called. Basically, it\\\'s all digital. So you take the photo, it scans your face like the banks use. So there are ways to do this completely securely and never have to share. And actually you. As the, as the, as the. As the landlord of the property manager doesn\\\'t even need to see those documents. Because they\\\'re all validated by a third party. And you\\\'ll pass. Yes, this is okay. And that\\\'s it. That\\\'s all you need to know.

Annabel Kaye: Absolutely. And I think one of the reasons they don\\\'t do that is they worry it\\\'s really expensive. Yeah, but nothing could be more expensive than emailing someone\\\'s passport to the other end of the planet. And them suffering data theft as a result. That wasn\\\'t much of an economy drive. But also to me, it makes people look dodgy, you know, if they can\\\'t answer that question.

Dr Steve Day: Anyway, you\\\'re talking, you\\\'re talking to a landlord that. When I started out, this is in my early 20s, I used to meet my tenants in the pub. And hand over contracts over the table and like, literally deposits came in, envelopes full of cash. And I thought this was totally fine and normal. Like you don\\\'t know what you don\\\'t know. So I completely understand when people do anything like this without. Without ever seeing that. Actually that could be perceived as not quite as legit as you should, because I was totally honest. I wasn\\\'t doing anything blowboard at all. Everything was above board is what I\\\'m trying to say.

But still, I could imagine looking back, I was like, wow. And the pubs we used to do it in. Because where my houses are in Manchester, they weren\\\'t the nicest pubs in the world. So this probably could have easily had some problems from, you know, that was.

Annabel Kaye: Probably before you had to look at their passport or their ID.

Dr Steve Day: Oh, way before any of that. I think we need to take photocopies. Anyway, we digress. There was a point and we are. We jumped onto a wonderful tangent then. Which is actually very, very interesting. So I don\\\'t want to stop you, but one of the point I was trying to make before was the practical aspects of this. And actually just thinking about when I\\\'m hiring a virtual assistant, I mentioned the data sharing agreement that we\\\'ve used. The copyright assignment agreement that we\\\'ve used in the past.

And I think that what I\\\'m hearing from now is be useful to have this additional step. Which would be in the terms of engagement or the hiring agreement. I imagine would be about, actually, that this person should be in all cases sharing their working. To actually just to basically cover my, my own back is you. Because what we do is we have a shared drive set up in our. We use Google, so our Google share drive and any work someone does for us.

We don\\\'t just want to see the end product in there, we want them to have their working documents in. We\\\'ve always done this, but actually it\\\'s never really been enforced. And I think from now on, I think that\\\'s going to be actually something that\\\'s going to come into our terms. It\\\'s like actually from the first draft, everything has to be saved in that drive. And these are the reasons why. And if you can\\\'t deal with that, then we don\\\'t want to know.

Or at least, for example, we have a team GPT account so people can actually log in. We store all of their chat history so we can actually see that kind of thing going on. And the videos that they record or the, you know, whatever they use. We have team accounts for that reason. So we\\\'ve got that history of everything that\\\'s done. So it just made me.

Annabel Kaye: At the same time, when it comes to personal data. Although you want it all stored. Ahat you don\\\'t want is everyone to have access to all the data. Yeah, you see what I mean? Which you go from what I call storing everything, which can turn into data hoarding. Right. But also sharing everything, which can be really insecure because you want layers of who has access to what. So you might want a couple of people in an organization, well, to see anything. You might want other people who just want to see what their team\\\'s up to. But when it comes to sensitive personal data share by your customers. Like tenant stage or whatever. You might not want lots of people to see that.

Because the more people see a passport, the more risk there is. And I know what entrepreneurs say to me. But I wouldn\\\'t hire people I don\\\'t trust. And it\\\'s like, well, of course not. But that doesn\\\'t mean you shouldn\\\'t secure the data. It\\\'s not about trust, it\\\'s about accidents. Right. If I lose my laptop and it\\\'s open and not encrypted. And I\\\'ve got access to everything in your system. Risk to you is much higher than if I lose it. And I\\\'ve got access to a subset of your system, doesn\\\'t mean I\\\'m not trustworthy. It means accidents happen and we\\\'re all people. Because presumably you wouldn\\\'t hire anyone. You didn\\\'t Trust, wouldn\\\'t it hire?

Dr Steve Day: Yeah, but also. I\\\'m sorry, I just think that\\\'s a bit of a naive way of looking at the world. That you will hopefully never hire anyone that you shouldn\\\'t trust. How on earth when you hire somebody you\\\'ve never met, can you trust them? You can think they\\\'re a nice person, but you can\\\'t know. And I think.

Annabel Kaye: But even if they\\\'re a nice person, they could be an efficient person and very nice. They could be sloppy about it, security and very nice. And they may be morally trustworthy, but technically delinquent. I\\\'ve met a lot of those.

Dr Steve Day: Totally. Or they could become over as incredibly nice. And they could come over as they\\\'re being very moral and putting a. Probably like a. What\\\'s the word I\\\'m looking for? A looking at the other side of the fence. If you like. They could actually be an evil person coming to try and rob your data and sell it. And I think you\\\'ve just got to. So if you\\\'re just being naive in saying I would never hire someone who\\\'s not trustworthy. I think that\\\'s just looking through the world to roast into glasses. Of course we wouldn\\\'t want to do that, but we can\\\'t all.

Annabel Kaye: Well, we\\\'ve all done it even if we didn\\\'t want to. And the difference between that and disaster was what they had access to. And how fast we knew that something was going wrong.

Dr Steve Day: I agree.

Annabel Kaye: But yeah, I think you\\\'ve got to trust people enough to hire them, but maybe not blind trust, isn\\\'t it? Yeah, but I think a lot of entrepreneurs are like that because for many people taking on. I think there\\\'s two categories of people taking on a VA. People who have run an employed team before which VAs aren\\\'t employed either in an executive position. And they know how reliable or not people are. And there are people who have never had a supervisory or team leading role. Who\\\'ve gone out and created something and if you like, their first hires are outsourced.

And in that sense they don\\\'t have the hinterland in their mind of what happens when you work with people. They just think, well, I asked them to do it, so that\\\'s what they\\\'ll do. And you don\\\'t have to have been a boss that long before you realize that\\\'s not the whole picture.

Dr Steve Day: No. And even with people with the best intentions and like you say. The total moral compass in the right direction, actually they just are either lazy. Or they dial attention to detail or they\\\'re busy and they forget something or they\\\'re tired or whatever it is. And it\\\'s too easy for things to be done. So when I worked in the hospital, I run a chair, I should say a committee called the Lessons Learned Committee.

And it\\\'s always about how to look at patient safety incidents, or in this case, data protection instance. In my case, it was patience and it was. It was basically looking at how do we improve the system to avoid this ever happening again. Because we can\\\'t trust the people because the people are human and humans make mistakes. And that is inevitable. And that\\\'s the only thing we can guarantee. Is the mistakes will happen again.

So how do you build a system better so that actually those mistakes don\\\'t have the impact? In this case, it did. And that\\\'s what really got me into the world of what I do now. And this is a great example of that. Is actually being mindful of the things, your protections and everything you\\\'re putting in place. The barriers you said to who has access to all the things you just mentioned. If that\\\'s the mindset you go into this, then it doesn\\\'t matter. It\\\'s not as important when something goes wrong. Because you\\\'ve done everything you can to minimize the impact to both you and the potential.

Annabel Kaye: And the only thing I would add into that statement. And that process now is be aware of the fact that AI can make mistakes. It can and it does hallucinate data. You need to watch your settings if you\\\'re doing data data analytics on anything confidential, personal. So it\\\'s not feeding it back into the general learning pool.

Dr Steve Day: Yeah.

Annabel Kaye: It\\\'s not infallible. I have vows with ChatGPT all the time. And I go, what the hell? That\\\'s absolutely wrong and you know it. And it\\\'ll go back. And I\\\'ve checked it again. You\\\'re right. I even got it to apologize to me the other day. Because I had to make it go back three times.

Dr Steve Day: I might apologize all the time. He says, oh, yes, thank you. All right. Thanks for missing that out. Sorry for the oversight. That\\\'s one of my favorite comments. Sorry for the oversight. I\\\'m like, anyway.

Annabel Kaye: I don\\\'t have any VAs, by the way, or as absent mind is as ChatGPT is. I don\\\'t think it\\\'s ready to take over the world. It\\\'s great at crunching data really quickly. But it actually does run out of memory and miss bits, doesn\\\'t it?

Dr Steve Day: Yeah. And as you said, it just makes stuff up. It just wants to please you. Yeah.

Annabel Kaye: Say anything you like? It\\\'s like. I always think it\\\'s like that boyfriend, when you say, does my mum look big? And they say, always goes, no, of course not. You look wonderful. And then you wear that frock and you think, someone should have told me, you know. I\\\'d prefer the one that went, well, this really isn\\\'t the outfit.

Dr Steve Day: For you or I actually, I really like that black one again.

Annabel Kaye: Yeah, but it is it. I don\\\'t. When they put the personality in it, they gave me the personality of a spaniel for some reason. Always wagging its tail and trying to sit on your feet. And yeah, it has its limits. And so I would say the same that you need to put in systems to think. Do I want to just rely on that as if it was like the Oracle, Adelphi and. Or am I actually going to have checks and balances on what it produces and what I accept from it?

I mean, a number of times I\\\'ve had to say to you, go back and do that again. And take out the following sources. Which are nowhere reliable sources of information on anything. Not even the weather, never mind what we\\\'re chatting about. And then it\\\'ll come back with a different point of view. It\\\'s like having a willing child who does their homework on drugs. I think, fantastic. I get out of it, maybe I\\\'m rubbish, but I\\\'m always arguing with it. My husband will go, what are you arguing with the computer about?

And I go, well, that\\\'s just wrong, you know, I don\\\'t mean wrong as a matter of opinion. But, you know, today it\\\'s Thursday. What do you mean? You\\\'re telling me it\\\'s Friday? You know, that kind of wrong, but.

Dr Steve Day: No, I. I completely agree. I think, yeah, take everything with a pinch of salt.

Annabel Kaye: So, you know, we build systems, don\\\'t we? And within that we build paperwork to require the people we work with to follow sensible systems. And you could put the whole across border data privacy under that heading, right, if you kind of got it. Because for all the jargon and all the language and all the fuss. That\\\'s ultimately what we\\\'re asked to do. And we don\\\'t even have to name it in weird bureaucratic language. As long as we comply with it. But to go stuff and nonsense, you know, who cares is otherwise. It makes an unstable business as you grow, doesn\\\'t it?

Dr Steve Day: Yeah, it does. Annabel, I\\\'m gonna have to draw this one to a close. Although I could chat with you for hours. And I always enjoy our conversations.

Annabel Kaye: I\\\'ll try one day to write an email you like again. But I haven\\\'t met one since, by the way, because I hate feel people\\\'s inboxes with loads of blah. I only write an email when I feel I\\\'ve got something useful to say.

Dr Steve Day: Yeah, I think that\\\'s a. That\\\'s a very good strategy. That\\\'s one I\\\'ve tried to do over the years as well. Which I only recall the podcast when I got something hopefully interesting to say. Otherwise if I can\\\'t think of anything, then I interview someone interesting like you. And I can ask you all the questions instead. So, thank you very much.

Annabel Kaye: Or we could get ChatGPT to go. Lessons have been learned.

Dr Steve Day: Exactly. Yeah, I did do it. Yeah, I\\\'m not actually, not done an interview with ChatGPT on live yet. That\\\'s a good one. I\\\'ll try that one. Yeah.

Fantastic. Annabel, thank you so much for coming on the show today. Just in case, if anyone wants to find out how Annabel can help with contracts. With legal advice on all this kind of stuff we\\\'ve been talking about. Then head over to sys.academy/tools, and you\\\'ll find a link on there. And you get over discount using our affiliate link as well.

And finally, please do subscribe. Hit subscribe right now. Share this with your friends and loved ones. Anyone who, you know, who runs a business you think would find it interesting. To know a little bit more about AI and copyright. And in my opinion, that should be pretty much every business owner you know.

Annabel, thank you very much indeed. It\\\'s been a pleasure. I\\\'ll see you next time.

Annabel Kaye: See you soon. Bye.

Dr Steve Day: Bye.

The Hidden Copyright Risk in AI-Generated Content—and How to Protect Your Business with Annabel Kaye, Founder of KoffeeKlatch | Ep 234

In this deep dive interview, we get into the weeds on:

KEY TAKEAWAYS: What You Need to Know About AI Copyright Law

BEST MOMENTS: Real Talk and Revelations About AI Copyright Law

TIMESTAMPED OVERVIEW: AI Copyright Law—Protect Your Business with Annabel Kaye

Episode Transcript

VALUABLE RESOURCES

LINKS TO CONNECT WITH THE GUEST

ABOUT THE GUEST

LINKS TO CONNECT WITH THE HOST

ABOUT THE HOST

For more articles related to legal considerations when outsourcing work and generating content using AI, you may also like:

SYS 163: Pro Tips for Calendar & Task Management to Optimize Your Productivity

SYS 144: Our Proven Management Toolkit is the Secret to why our Virtual Assistants Stick Around for the Long-term (average 36 months and counting!)

SYS 121: The Quick and Easy Guide to Ethically Recreating Top Influencers’ Marketing Funnels Using AI Tools (and a few cool tricks!)

Discover How Small Business Owners Leverage The SixR System To Achieve Time Freedom In Under 90 Days

How Small Business Owners
Leverage The SixR System
To Achieve Time Freedom
In Under 90 Days

Free On-Demand Masterclass