Latest / #ContentChat / 3/30/26 #ContentChat: AI Copyright Risks for Content Marketers

Transcript

3/30/26 #ContentChat: AI Copyright Risks for Content Marketers · #ContentChat

• 0:01
  SPEAKER_00: Good afternoon, everybody.
• 0:02
  SPEAKER_00: I'm Erica Heald, your content chat host.
• 0:05
  SPEAKER_00: And joining me today is my friend Karache Gorgoni.
• 0:09
  SPEAKER_00: Oh God, I forgot to ask you if I was doing your last name right.
• 0:12
  SPEAKER_00: I hate it when I do that.
• 0:13
  SPEAKER_00: Oh my god.
• 0:14
  SPEAKER_01: I mean, you're not, but it's okay.
• 0:16
  SPEAKER_00: It's like when people tell me Erica Held, and I go, Oh crud.
• 0:20
  SPEAKER_01: It's Karache Gorgon, but it, you know, it depends like what generation of folks you ask.
• 0:27
  SPEAKER_01: Because my husband's grandfather said Gorgoni, but over here we just discovered that like telemarketers will add an I.
• 0:33
  SPEAKER_01: And we don't, we it's not correct.
• 0:36
  SPEAKER_01: We don't say it like that.
• 0:41
  SPEAKER_00: Well, today we're talking about an issue where these kind of things actually arise all the time.
• 0:46
  SPEAKER_00: Because when you're using AI, there's all sorts of things that happen.
• 0:49
  SPEAKER_00: And you got to be prepared for them.
• 0:51
  SPEAKER_00: And you need to understand even what all the things are to prepare for.
• 0:55
  SPEAKER_00: And so I know that's a lot to promise to pack into an hour-ish, but I think we're going to try pretty darn hard.
• 1:02
  SPEAKER_00: But first, uh, Carrie, can you take a minute to introduce yourself properly and tell folks a little bit about who you are, what you do, and why this is like your superpower?
• 1:11
  SPEAKER_01: I don't know about properly.
• 1:12
  SPEAKER_01: Uh, but I'm Carrie O'Shea Gorgon, and I'm a recovering attorney.
• 1:17
  SPEAKER_01: So I graduated first in my law school class, practiced for about five years, then got into teaching, then got into training and speaking at conferences, and now I do marketing.
• 1:29
  SPEAKER_01: Um, I was with marketing profs for about eight years.
• 1:32
  SPEAKER_01: Some of your audience might know those folks.
• 1:34
  SPEAKER_01: And now I'm with a company called App Fire, and I'm on the content team.
• 1:38
  SPEAKER_01: And uh one of my superpowers is using AI to speed along our workflows.
• 1:43
  SPEAKER_01: So I'm intimately familiar with how AI can expedite things and how it can make things uh blow up in your face.
• 1:51
  SPEAKER_00: I think your team is so lucky to have you because I can't imagine how awesome it would be.
• 1:57
  SPEAKER_00: Because I know the conversations I would get into with, um, especially in a startup or bootstrapped kind of um founder situation where folks don't really understand how AI affects things like your copyright.
• 2:12
  SPEAKER_00: They think, you know, oh well, everyone's using it, everyone's doing the thing, so it's fine, right?
• 2:17
  SPEAKER_00: We can do whatever we want.
• 2:18
  SPEAKER_00: And it's like, uh no.
• 2:22
  SPEAKER_01: There's there's so much.
• 2:23
  SPEAKER_01: There's so much.
• 2:24
  SPEAKER_01: I mean, for one thing, I started to get nervous about uh folks kind of experimenting with custom GPTs, which is like now it's like old hat, right?
• 2:34
  SPEAKER_01: But loading stuff into it that we didn't create.
• 2:38
  SPEAKER_01: So if it's not ours, I am very hesitant to just train my little agent on somebody else's, say, custom research or something.
• 2:49
  SPEAKER_01: Is it going to spit it back out again such that now you're violating somebody's copyright?
• 2:53
  SPEAKER_01: Like, no, probably not.
• 2:54
  SPEAKER_01: I mean, if it's like say SEO research and you put it into your custom GPT and then you say, I'm gonna write a blog post and I want you to analyze my blog post for SEO stuff, like you're not reposting their SEO stuff anywhere.
• 3:09
  SPEAKER_01: You're actually applying it.
• 3:10
  SPEAKER_01: So it depends how you look at it, whether or not that would be considered fair use.
• 3:14
  SPEAKER_01: Like you're applying the knowledge.
• 3:16
  SPEAKER_01: I could do that from reading their book, or I can put it in my little custom GPT, but it is not clear by any stretch of the imagination yet if that's cool or not cool, like from a legal standpoint.
• 3:27
  SPEAKER_00: Yeah.
• 3:27
  SPEAKER_00: And I think a lot of companies are just now even starting to think about putting some of those um clauses about what is fair use for their um for their intellectual property.
• 3:40
  SPEAKER_00: I have I've had exactly one client in the last four years who has um had language around that because they are a large um management training um company.
• 3:53
  SPEAKER_00: So their business is their content and their intellectual property.
• 3:57
  SPEAKER_00: And so they make it clear in every license that they have and on their website that all of their information is they they own it, and that you are not authorized to upload any of their copyrighted material into a GPT.
• 4:12
  SPEAKER_00: But how are they going to know?
• 4:14
  SPEAKER_01: Well, yeah, I mean, that's part of the problem.
• 4:16
  SPEAKER_01: I think in any case, what they say cannot change the parameters of fair use, they may not be happy about it.
• 4:24
  SPEAKER_01: But it's true that in general, fair use has gotten like smaller and smaller and smaller, the uh the kind of universe of stuff that you can use and what purpose you're using it for.
• 4:35
  SPEAKER_01: It's sort of like how in documentaries in the 1970s, like they might pick up a song that's playing on the street or a little bit of like TVs posted over a bar or something, and it was not a big deal.
• 4:46
  SPEAKER_01: Well, it's a big deal now because companies will write them, they will make demand letters and stuff.
• 4:50
  SPEAKER_01: And most people, they just don't want the headache.
• 4:52
  SPEAKER_01: So even though if you litigated it and took it all the way through to the end, it might be fair use.
• 4:57
  SPEAKER_01: Most people, they just can't be bothered.
• 4:59
  SPEAKER_01: So it's uh that's kind of like the environment we're operating in.
• 5:03
  SPEAKER_01: But I think you're right.
• 5:04
  SPEAKER_01: If you're using your custom GPT or really any, you could use Claude, you could use uh Claude Skills, you could use any AI like flavor of the week that you like the best.
• 5:15
  SPEAKER_01: If you're putting stuff into it that you don't own, what you're doing with it is gonna determine whether or not that use is fair or not.
• 5:23
  SPEAKER_01: I mean, it's not clear yet.
• 5:25
  SPEAKER_01: So I think your risk factor, kind of like your risk tolerance for yourself, has just got to be different than your risk tolerance for a company or client because you can't afford to take risks somebody else is gonna have to pay for.
• 5:39
  SPEAKER_01: So that's kind of how I look at it.
• 5:41
  SPEAKER_00: 100%.
• 5:42
  SPEAKER_00: And in most companies, it's the marketing team that is really taking those risks on behalf of the brand and frequently not understanding that they're taking a risk when they're using an AI tool and not mentioning it or even mentioning it, but not understanding how much of the how much of the human input is necessary for something to actually be IP that the brand can then claim ownership of later.
• 6:11
  SPEAKER_01: Yeah, oh, it's wild.
• 6:12
  SPEAKER_01: So there's so much to this.
• 6:14
  SPEAKER_01: For one thing, you can't in the United States at least claim copyright protection on anything that's AI generated, not your graphics, not your written content, no AI generated nothing.
• 6:26
  SPEAKER_01: So if copyright is a concern of yours, like let's say you make training content, you know, obviously it's the product, you need to be extra worried about it.
• 6:35
  SPEAKER_01: Um, then you need to be to be thinking about this pretty seriously and creating the content yourself and then using AI like as an assist, maybe for some research or maybe for some um pointing out typos or something like that.
• 6:47
  SPEAKER_01: But you're gonna want to take a light hand versus if you're using it to sell widgets, do you care if you own the blog post about widgets or not?
• 6:55
  SPEAKER_01: I mean, it's a good question.
• 6:57
  SPEAKER_01: You're in the business of selling widgets, you're not in the business of selling blog posts.
• 7:00
  SPEAKER_00: So that's if you're in the entertainment industry, if you're in the media, there's so many industries where people feel like, oh, well, I'm gonna use this tool to create the images because my my employer is cheap and won't actually get me, you know, a dedicated graphic designer or even a freelancer.
• 7:21
  SPEAKER_00: So I don't have any way to have images.
• 7:24
  SPEAKER_00: But then you end up in that situation where you don't own that amazing image that you create.
• 7:29
  SPEAKER_00: So it can be used however somebody who doesn't like you wants to, or worse, you're using somebody else's without realizing it.
• 7:36
  SPEAKER_01: Like that's happened so much that if you ask AI for something like um a red-haired woman holding a cup of coffee in her kitchen, it'll just find something kind of like that and give you something very similar.
• 7:49
  SPEAKER_01: Now that came from somewhere, it was trained on something.
• 7:52
  SPEAKER_01: So next thing you know, you're getting sued for like recreating somebody's book cover or something, and it's you definitely didn't mean to.
• 7:58
  SPEAKER_00: Yeah.
• 7:59
  SPEAKER_00: So that's a risk.
• 8:00
  SPEAKER_00: And that's not a legal defense, right?
• 8:02
  SPEAKER_01: I didn't know, obviously, any more than it would be a legal defense if you had like a subcontractor from Fiverr do it for you and they stole.
• 8:12
  SPEAKER_01: It's still you're the one who put it out there.
• 8:14
  SPEAKER_01: So it's it's a problem you're gonna have to contend with.
• 8:18
  SPEAKER_01: But the thing is, we've all been doing it ourselves for so long.
• 8:21
  SPEAKER_01: Like we didn't use AI, we didn't have the option, didn't know it was gonna be an option until pretty recently, right?
• 8:26
  SPEAKER_01: And I don't know about you, but I've been pumping out content for years just non-stop, like a hamster on a wheel.
• 8:33
  SPEAKER_01: So if everybody else's company is like that, why would you ever be like, I don't know, scrape Google for me?
• 8:38
  SPEAKER_01: You've got so much, just load it up with your own stuff.
• 8:42
  SPEAKER_01: You're gonna get a better result anyway.
• 8:44
  SPEAKER_00: Totally.
• 8:44
  SPEAKER_00: And you can do things like take your own terrible photos that you took of something, and you can, you know, put it into an AI tool and have them fix them for you so that way you know for sure, since the base material was yours, that it is something authentically not being stolen from someone else.
• 9:04
  SPEAKER_01: I mean, I always have in place some guardrails, like I will tell it specifically do not recreate anybody else's copyrighted material, do not use their direct language, do not, you know, just don't.
• 9:17
  SPEAKER_01: And if you think something maybe came from somewhere, give me a citation.
• 9:21
  SPEAKER_01: I'm gonna go check it out myself.
• 9:23
  SPEAKER_01: So let's not do any of that, let's not take those chances.
• 9:27
  SPEAKER_01: But I think the best use case I've seen for visual things in AI, it's not creating assets wholesale, it's creating something that can help designers figure out what it is you think you want because that's the hardest part filling out a creative brief.
• 9:42
  SPEAKER_01: I mean, some people are just insanely, they don't mean to be vague, but like it is it'd be like, I don't know, I want it to be like cotton candy, but Christmas.
• 9:50
  SPEAKER_01: And you're like, I don't know what that is, you know.
• 9:52
  SPEAKER_00: So you can show them just jazz this up, and you're like, what does that even mean?
• 9:59
  SPEAKER_01: Right, right.
• 10:00
  SPEAKER_01: Now, sometimes they don't know.
• 10:02
  SPEAKER_01: So if they don't know, then AI can't help them with that.
• 10:04
  SPEAKER_01: But if they're like, Yeah, I'm thinking something like this, this, and this, it can spit out a few different versions.
• 10:09
  SPEAKER_01: Then a design team could look at that and go, Oh, okay, I see what you mean.
• 10:12
  SPEAKER_01: Like this is sort of a jazz hands type of thing.
• 10:16
  SPEAKER_01: Gotcha.
• 10:17
  SPEAKER_01: Um, and I find it really, really helpful for that.
• 10:20
  SPEAKER_01: And you're not using the the output directly, you're just giving the design team that you continue to pay, for God's sake, please pay them some idea what you're talking about.
• 10:31
  SPEAKER_00: Totally.
• 10:33
  SPEAKER_00: So um I always rely on smart people like you and going to conferences to help me know what's going on.
• 10:40
  SPEAKER_00: Um, and I'm wondering if there have been any recent fair use related or copyright and AI-related uh lawsuits that folks should be aware of that are kind of um things that are just floating around there that we need to be thinking about.
• 10:55
  SPEAKER_00: Because it feels like the last big wave of them resulted in a lot of um media companies striking um paid deals with a lot of the AI companies that had already stolen their content and used it to train their models.
• 11:11
  SPEAKER_01: Yeah, I mean, that's true.
• 11:12
  SPEAKER_01: So Anthropic, I I love Claude, Claude's probably my favorite, but Anthropic is in kind of the middle of a pretty big settlement with people whose stuff they used um to train their AI without permission.
• 11:27
  SPEAKER_01: So, authors, if this is actually, I think, the last day to file your claim in that class action lawsuit.
• 11:34
  SPEAKER_01: If you have not, you're gonna want to do that.
• 11:36
  SPEAKER_01: Let me figure out what the uh what the actual URL is.
• 11:39
  SPEAKER_01: Anthropic copyright settlement.com.
• 11:42
  SPEAKER_01: You're gonna want to get on that today because the opt-out window already closed.
• 11:45
  SPEAKER_01: If you want to file your claim, you're gonna go do that.
• 11:48
  SPEAKER_01: But basically, they trained their AI on a bunch of people's copyrighted books, did not get permission.
• 11:52
  SPEAKER_01: So now they're they've got a settlement on the table and they've got uh thousands of authors filing claims against it.
• 12:00
  SPEAKER_01: Then I think at the end of April there'll be a hearing where the court decides whether that's fair or not.
• 12:05
  SPEAKER_01: And then the authors could get payouts based on how many of their copyrighted works Anthropic used.
• 12:11
  SPEAKER_01: So that's a pretty big one.
• 12:12
  SPEAKER_01: Uh, but there's been tons.
• 12:14
  SPEAKER_01: Like I think you probably saw Sora 2 just stopped operating.
• 12:18
  SPEAKER_01: Nothing but problems.
• 12:20
  SPEAKER_00: After making that big deal with Disney with Sora 2, yeah.
• 12:23
  SPEAKER_01: So that there's so much wrong with that.
• 12:26
  SPEAKER_01: Um, part of it was that they gave people the mechanism to to steal other people's copyrighted stuff to use their likeness for anything at all, including things that were inappropriate or things that were commercial in nature.
• 12:43
  SPEAKER_01: And they just went ahead and kind of abdicated responsibility for what people did with that technology.
• 12:49
  SPEAKER_01: And you just can't do that.
• 12:50
  SPEAKER_01: So I think they rightly realized in the end that it wasn't worth it because you're just never going to be able to close the door completely on um on infringing or wrongful and very expensive uses.
• 13:04
  SPEAKER_01: So yeah, I mean, things like people would take images of Stephen Hawking and doing horrific things, and uh his estate would obviously get upset for good reason.
• 13:13
  SPEAKER_01: So you just yeah, some stuff is just like never gonna be.
• 13:25
  SPEAKER_01: Let's see, nine books, ten books.
• 13:29
  SPEAKER_01: It's like gotta be at least a few thousand a book.
• 13:31
  SPEAKER_01: I have to think it's a good amount of money.
• 13:33
  SPEAKER_01: See, it's a good reason to start pumping out those books.
• 13:35
  SPEAKER_01: Did they use all nine though?
• 13:36
  SPEAKER_01: That's a good question.
• 13:37
  SPEAKER_01: We'll have to ask them that because I'm pretty sure there was one in there, like it's not about the tights.
• 13:41
  SPEAKER_01: That was about what is that about being a superhero.
• 13:43
  SPEAKER_01: I don't think they picked that one up somehow.
• 13:46
  SPEAKER_00: Well, that's funny.
• 13:47
  SPEAKER_01: I'm just kidding.
• 13:50
  SPEAKER_01: Let's give them a hard time with that.
• 13:51
  SPEAKER_01: But if you've written a book and Anthropic used it to train their AI, get on that, you get a file today, or you're gonna miss the window.
• 13:59
  SPEAKER_01: Get your settlement money.
• 14:00
  SPEAKER_01: Um, what other kind of stuff?
• 14:02
  SPEAKER_01: Let's see.
• 14:02
  SPEAKER_01: So we've actually just talked about the fact that you can't copyright anything that was AI created.
• 14:08
  SPEAKER_01: You're gonna have to prove that you made a significant human contribution to that work.
• 14:13
  SPEAKER_01: And frankly, that would be your best bet anyway, because the AI, I can't even say this, the EU AI Act uh is a lot more stringent than anything we've got in the US right now.
• 14:25
  SPEAKER_01: For one thing, we have this weird like patchwork of laws at the state and federal level.
• 14:30
  SPEAKER_01: It's kind of hard to follow.
• 14:32
  SPEAKER_01: But if you look at Europe, that's always where we end up.
• 14:35
  SPEAKER_01: Because if, like me, you work in B2B SaaS, you definitely have some clients in Europe.
• 14:41
  SPEAKER_01: And as long as you do have customers or clients in Europe, you're going to have to comply with the EU AI Act regardless.
• 14:48
  SPEAKER_01: So it's just a good practice to start saving all of your conversations right now and to keep receipts kind of for how much human contribution you put into the content that you're creating using AI that will help.
• 15:03
  SPEAKER_01: It does not like I can't say for sure that it's going to save your bacon someday, but it will get you at least part of the way to being able to defend yourself against a copyright lawsuit involving AI-generated content.
• 15:15
  SPEAKER_01: So I put together right before this, I was like, hey, I wonder if anybody's ever, you know, put together a template that people could use for uh tracking how much human input there was for their content versus how much AI.
• 15:27
  SPEAKER_01: So I put something together for that.
• 15:29
  SPEAKER_01: You can share that link.
• 15:30
  SPEAKER_01: It's seriously, don't get like really excited.
• 15:32
  SPEAKER_01: It's a Google Doc on purpose.
• 15:34
  SPEAKER_01: It's quick and dirty so that you can copy it and uh and just make it a work file, basically.
• 15:40
  SPEAKER_01: For every big piece of content you create, you can make this work file to go with it that will help document uh the fact that say you started the process with your own outline or your own notes and then turned it into something else versus asking AI to write it for you from scratch, which would be a very different equation.
• 15:58
  SPEAKER_00: And it's not to say that that's never appropriate.
• 16:01
  SPEAKER_00: If you're doing like a Facebook ad for something that's not really important, you're scrappy, go ahead, use AI to create that.
• 16:09
  SPEAKER_00: But you need to make sure that anything that's gonna be value and be your intellectual property down the line, well, it has to be your intellectual property.
• 16:18
  SPEAKER_00: You have to be the one that created it with AI just as your little helper behind the scenes.
• 16:23
  SPEAKER_01: Yes.
• 16:24
  SPEAKER_01: And make sure that you are not copying anyone else because there is absolutely no way your Facebook ad that you put out is gonna make you more than it will cost you in copyright infringement damages if you stole from someone else by mistake.
• 16:38
  SPEAKER_00: That's why you need to steal from yourself.
• 16:40
  SPEAKER_00: That's what I really like to plagiarize myself.
• 16:43
  SPEAKER_00: I upload all of my own things into my GPT so that way it knows, you know, here's what I've written, here are transcripts from my presentations, here are my presentation decks.
• 16:56
  SPEAKER_00: I'm like, steal from me.
• 16:58
  SPEAKER_00: And I'm always let more people don't do that.
• 17:01
  SPEAKER_01: So my favorite way to use AI is to take something I've created, put it into whatever my my favorite one is, Claude, but you could use whatever you want.
• 17:09
  SPEAKER_01: And I'll be like, hey, you are now whatever whoever is going to be reviewing this.
• 17:14
  SPEAKER_01: I want you to grade this on a scale from F being absolute garbage start over, to A plus being let's build and and deploy this today.
• 17:22
  SPEAKER_01: Give me detailed feedback.
• 17:24
  SPEAKER_01: You know, here's whatever brief I worked off of, here's the uh the task specifications.
• 17:29
  SPEAKER_01: Let me know how I did.
• 17:30
  SPEAKER_01: And it will give you a pretty good list of things to refine and make it better for you.
• 17:36
  SPEAKER_01: And meanwhile, the content I put into it initially, it's all mine.
• 17:39
  SPEAKER_00: Yep.
• 17:40
  SPEAKER_00: It's a really good coach because it'll tell you, hey, you're close here, but for the voice and tone you're going for, for your audience, you know, this isn't quite right, or hey, these things are a little bit repetitive.
• 17:51
  SPEAKER_00: I love it for that because you know, we used to have more humans in our life who had those roles in our organizations.
• 17:58
  SPEAKER_00: Um, but like I I'm, you know, a per solopreneur, it's just me.
• 18:02
  SPEAKER_00: Um, so while you know, I do have folks I try to do.
• 18:06
  SPEAKER_00: It's lonely.
• 18:07
  SPEAKER_00: So I'm not saying that these are my friends because you know, I do have I have Kupo here.
• 18:11
  SPEAKER_00: He's my friend, but he can't read and he can't really provide me with any uh meaningful feedback on anything.
• 18:17
  SPEAKER_01: Dog illiteracy is a huge problem.
• 18:20
  SPEAKER_00: It really is.
• 18:20
  SPEAKER_00: We were talking to our other bulldog about it last night, too.
• 18:23
  SPEAKER_00: Like he he really gets sad he can't read.
• 18:28
  SPEAKER_01: Well, my friend, our friend who came by in the comments, Chris Brogan, makes fun of me for trying to get that A plus from my AI reviewer.
• 18:37
  SPEAKER_00: But I why wouldn't you?
• 18:39
  SPEAKER_00: Woman who finished number one in her law class likes being graded.
• 18:43
  SPEAKER_01: Yep.
• 18:44
  SPEAKER_01: I just if you can get an A plus, like why wouldn't you?
• 18:47
  SPEAKER_01: Yeah, that's the way I look at it.
• 18:49
  SPEAKER_01: But you've got to be careful too to coach your AI not to kiss your butt because it's like, I don't know, they're all kind of programmed to want to tell you that you are correct and wonderful and smart.
• 19:01
  SPEAKER_01: So I actually took a prompt that says something like, Hey, in my culture, it's offensive to give people compliments for no reason.
• 19:08
  SPEAKER_01: So stop doing it immediately and talk to me appropriately as though you know we were like professional colleagues and stop kissing my butt.
• 19:20
  SPEAKER_00: But that's also why it's so much more useful to be building some of these projects or tools versus just starting from a prompt, because when you just start from a prompt, you lack all of that context and nuance that you can build in when you are really treating it as an operating system and a project management tool and giving it all the things that it needs, um, especially with all the don'ts.
• 19:43
  SPEAKER_01: Well, now there's so much to think about here.
• 19:46
  SPEAKER_01: Um, I look at it, I'm we're talking about it from a copyright standpoint, but just from an operational standpoint, it's really come to the point in its development AI has that you could supervise a whole team of AI agents to do an entire campaign or something for you.
• 20:02
  SPEAKER_01: So you could have 12, 13 working at once on different pieces of your campaign with one orchestrator agent running them all.
• 20:10
  SPEAKER_01: So the thing is you've got to make sure that you've built guardrails, built compliance into every one of those agents to the extent you're able, and that at the end of it, you are going to take a real and meaningful look at everything that came out the other side to make sure that it's not infringing, that it doesn't look or sound like perhaps it came from someone else.
• 20:29
  SPEAKER_01: Or worse, like if it doesn't sound like anything you put in, where did it come from?
• 20:34
  SPEAKER_01: Right.
• 20:34
  SPEAKER_01: And then you know you're gonna want to look at it anyway, aren't you?
• 20:36
  SPEAKER_01: To make sure that it's actually your like what you wanted.
• 20:40
  SPEAKER_01: So you could, in theory, turn on all the switches and let it like post your stuff for you without a human review.
• 20:46
  SPEAKER_01: I would never ever recommend it.
• 20:49
  SPEAKER_00: But that's the problem with how everyone is promoting AI adoption as a way to be more efficient, to save time, to have fewer humans.
• 21:00
  SPEAKER_00: But the thing is, is this is also how to get sued more.
• 21:04
  SPEAKER_01: Yeah, I mean, potentially.
• 21:06
  SPEAKER_01: So I I wouldn't say that it's um impossible to have fewer humans.
• 21:11
  SPEAKER_01: I think you could, but the humans that you keep are going to have to be the ones with the right expertise to look at what's coming out the other side, not just for quality, which you absolutely need to look at, but for compliance, which is a whole nother ball of wax.
• 21:27
  SPEAKER_01: Most people are not thinking about compliance.
• 21:30
  SPEAKER_01: They think that because a tool can do a thing, that that thing must be legal.
• 21:35
  SPEAKER_01: It is not true.
• 21:36
  SPEAKER_01: It is not true.
• 21:37
  SPEAKER_01: I used to tell people this.
• 21:38
  SPEAKER_01: Do you remember rafflecopter?
• 21:39
  SPEAKER_01: Like back in the day.
• 21:40
  SPEAKER_01: Yes.
• 21:41
  SPEAKER_01: People were so excited they could do a raffle online using raffle copter.
• 21:46
  SPEAKER_01: Yes.
• 21:47
  SPEAKER_01: And I was like, that's not legal.
• 21:48
  SPEAKER_01: You're not, you're literally not allowed to win something by chance in Canada and in certain countries.
• 21:53
  SPEAKER_01: Like it's not legal to do it that way.
• 21:55
  SPEAKER_01: And they'd be like, but raffle copter lets me.
• 21:57
  SPEAKER_01: I'm like, well, raffle copter doesn't care about you.
• 22:00
  SPEAKER_01: Because raffle copter is not gonna have to pay for the lawsuit that you're gonna, you know, find yourself in the middle of.
• 22:05
  SPEAKER_01: Like tools don't care about you.
• 22:08
  SPEAKER_01: They have the capabilities they have.
• 22:10
  SPEAKER_01: It is up to you to have the judgment to use them.
• 22:13
  SPEAKER_00: And if we're giving people no training on our AI tools in the first place, how on earth can we expect them to know what to look at?
• 22:20
  SPEAKER_00: 100%.
• 22:21
  SPEAKER_00: Yes.
• 22:23
  SPEAKER_00: They should be putting in or not.
• 22:24
  SPEAKER_00: I mean, I don't see that people, the same people who had the poor judgment of having like spreadsheets with client data on laptops that they left unattended, uh, you know, in public places, are the same people using AI in our organizations today.
• 22:40
  SPEAKER_01: Isn't that scary?
• 22:42
  SPEAKER_01: So I feel like everybody right now who's wanting to just show everybody, like their investors or whatever, that they're using AI, they should take a step back because showing them that you're using is one thing, but you don't necessarily want to showcase all the risk you're putting on the table if you have not done the necessary kind of preliminary steps, governance, uh, leadership, training to make sure that people don't put client data into your AI, that they don't put any kind of personally identifiable information into AI, that they don't put you know any of your kind of financial stuff, like nothing you're gonna want showing up someday in a data breach.
• 23:25
  SPEAKER_01: You just can't put it in there.
• 23:27
  SPEAKER_01: I wish that wasn't the case, but even in your enterprise accounts that are supposed to be behind the firewall, even then there are certain categories of things you should just never ever put into AI.
• 23:39
  SPEAKER_00: Yeah, 100%.
• 23:40
  SPEAKER_01: And unless you're telling people that, they I mean, how would they know?
• 23:44
  SPEAKER_01: So you've gotta you've got to back it up, work with a company like Trust Insights.
• 23:49
  SPEAKER_01: And I don't have any like financial interest in Trust Insights with Christopher Penn and Katie Robert, but they start at that level up where it's like, let's look at your company, your goals, your workflows, and your systems, figure out where you can use AI responsibly and securely to help you, you know, achieve your goals faster without exposing yourself to liability.
• 24:09
  SPEAKER_01: So I'm like, yes, please, let's think about it that way first and not kind of do it the way a lot of companies are doing it, which is to just let these little fires crop up across the org.
• 24:21
  SPEAKER_00: So are you seeing any specific really common use cases that are feeling really risky right now?
• 24:28
  SPEAKER_00: Like things that you're seeing, um, where marketing teams are really exposing their companies to risk without realizing it.
• 24:36
  SPEAKER_01: There's a few different categories of risk.
• 24:38
  SPEAKER_01: And so uh I'd say one of them is people who use AI for things like financial analysis for reports they're going to give to the board or some kind of other very important stakeholder, and they're not checking it.
• 24:50
  SPEAKER_02: Yeah.
• 24:51
  SPEAKER_01: So I mean, just like AI, a lot of it's garbage at math.
• 24:54
  SPEAKER_01: It like just is.
• 24:55
  SPEAKER_01: It can make you the code to crunch the numbers, but it can't crunch the numbers for you always.
• 25:01
  SPEAKER_01: It's it's not safe to rely on that without checking.
• 25:04
  SPEAKER_01: So I would not use it for that.
• 25:06
  SPEAKER_01: I might use it to help me kind of make talking points around things.
• 25:10
  SPEAKER_01: Um, but again, only if the data is not super sensitive.
• 25:13
  SPEAKER_01: Yeah.
• 25:13
  SPEAKER_01: So that's one thing.
• 25:15
  SPEAKER_01: Keep an eye on using it for any kind of like math or financial analysis.
• 25:19
  SPEAKER_01: That's that's going to require like triple checking in my book.
• 25:22
  SPEAKER_01: Um, and then anything like marketing, we're we're hugely exposed because there's such a high pressure for reasons I still don't understand, but there's just this constant pressure to turn things around in insane amounts of time, like tiny little how long it takes to do the thing.
• 25:38
  SPEAKER_00: So they think it must be automatic, right?
• 25:40
  SPEAKER_00: So automatic.
• 25:42
  SPEAKER_01: So like we're gonna be in an event next week, so why can't you just do our entire, you know, presentation plus all the social posts to promote it, plus whatever now, and a new booth and maybe a few handouts.
• 25:54
  SPEAKER_01: Yeah.
• 25:54
  SPEAKER_01: So that high pressure turnaround time, it creates this pressure cooker in which people's judgment gets kind of turned off because they're panicked.
• 26:02
  SPEAKER_01: And that's not being panicked is not a great way to approach anything that requires strategic thinking, but especially not legal liability, right?
• 26:09
  SPEAKER_01: So unless you've built guardrails in before that, you you run a real risk of people shortcutting and you know, maybe taking somebody else's post and being like, just you know, make this more like us, or uh, you take this and make a blog post out of it.
• 26:24
  SPEAKER_01: Like, you don't ever want to create that kind of temptation, that kind of environment.
• 26:29
  SPEAKER_01: Um, but I've got blogs.
• 26:30
  SPEAKER_00: If you get leaders asking you to do that, you'll have people straight up handing you something and say, We need this.
• 26:37
  SPEAKER_01: You know, I have never had that experience.
• 26:39
  SPEAKER_01: Thank God.
• 26:40
  SPEAKER_01: Oh, I have not, but I'm really really sure you have.
• 26:43
  SPEAKER_00: And I've been like, No, we're not doing that.
• 26:46
  SPEAKER_00: Thanks.
• 26:47
  SPEAKER_01: But I actually put guardrails in place in the AIs, any kind of custom GPTs or agents I've created, but every conversation that I have, I'll say, You are absolutely not to lift language from somebody else's copyrighted material.
• 27:01
  SPEAKER_01: Do not, and not even like, don't just paraphrase it, like, don't take somebody else's ideas.
• 27:06
  SPEAKER_01: These are my ideas, use mine.
• 27:08
  SPEAKER_00: Yeah.
• 27:09
  SPEAKER_01: And then you know, you can make suggestions for where else I might look for inspiration, but tell me and I'll go look.
• 27:15
  SPEAKER_01: You've got to be so careful with that stuff.
• 27:17
  SPEAKER_01: So that's I mean, probably the most common use cases are to take um to create something completely out of whole cloth, which is the riskiest use case for AI, because without you telling it where to get that from, it's I mean, by definition, it's taking it from somewhere.
• 27:32
  SPEAKER_02: Yeah.
• 27:32
  SPEAKER_01: So you'd better give it your own thing.
• 27:35
  SPEAKER_01: So, like in the case of an event that's coming up, maybe it's an event about Agile, and you just did a white paper about Agile last quarter.
• 27:43
  SPEAKER_01: So grab that white paper, cram it in, and ask it to make some things for you, some presentation topics that you can run by your boss or some blog posts or some social posts.
• 27:52
  SPEAKER_01: Don't ask it to make it up out of nothing because that nothing is actually somebody else's copyrighted material.
• 27:58
  SPEAKER_01: Like full stop.
• 27:59
  SPEAKER_00: It goes back to the plagiarizing your own ideas, which is so important.
• 28:03
  SPEAKER_01: Yeah.
• 28:04
  SPEAKER_00: You know, but what you're also talking about is people, you know, marketing frequently is one of the most siloed groups within any organization.
• 28:13
  SPEAKER_00: Sometimes they're even literally physically shoved in a back room, you know, where they will have as the least amount of contact with the rest of the organization as possible.
• 28:23
  SPEAKER_00: And so folks in marketing may not even know who their their legal contact is if they have a large legal team.
• 28:31
  SPEAKER_00: They may not, you know, have experience with compliance.
• 28:34
  SPEAKER_00: I worked with Charles Schwab for 10 years.
• 28:36
  SPEAKER_00: So I had my own compliance person and I knew who the EVP was of legal, that things would be escalated to if there's any kind of an issue.
• 28:46
  SPEAKER_00: So I was able to bake them cookies and bring it to them and be in their face a lot so that they knew who I was, so that we could work together on things.
• 28:55
  SPEAKER_00: So we would start um at the beginning of any kind of big project that we knew was potentially going to be sensitive.
• 29:02
  SPEAKER_00: But we also put lots of things in place for risk management where we would have like crisis plans.
• 29:08
  SPEAKER_00: And I know that's something that um for your marketing prof session, you had um kind of an outline.
• 29:13
  SPEAKER_00: And I'd love it if we could just kind of spend a minute talking about the importance of one, having those legal relationships and two, crisis planning, because you're we're all know that we're gonna have an issue at some point.
• 29:26
  SPEAKER_00: We're not perfect, no matter how hard you try to be ahead of this.
• 29:29
  SPEAKER_00: There's gonna be a problem.
• 29:31
  SPEAKER_00: So how are you gonna react to it other than running around with your hair in fire, which is not good?
• 29:36
  SPEAKER_01: Well, the problem is by the time, all right, so a crisis hits now.
• 29:39
  SPEAKER_01: You're like, oh no, I should have stayed in touch with legal before now.
• 29:44
  SPEAKER_01: It's too late now.
• 29:45
  SPEAKER_01: Okay, so that's problem number one.
• 29:46
  SPEAKER_01: Problem number two is legal without some healthy interactions with marketing or other teams using AI, legal may not be completely up to speed on it either.
• 29:58
  SPEAKER_01: It is an incredibly fast-moving area of technology and law.
• 30:03
  SPEAKER_01: So the law that applies to AI, it's already exists.
• 30:07
  SPEAKER_01: People think it's like the Wild West and there's no law.
• 30:10
  SPEAKER_01: There is.
• 30:10
  SPEAKER_01: It exists, it just hasn't explicitly been, you know, applied to AI before because AI didn't exist before.
• 30:17
  SPEAKER_01: So, like copyright and uh privacy law and financial related laws, like they they all these things exist.
• 30:26
  SPEAKER_01: Keep that in mind too.
• 30:28
  SPEAKER_01: But law, legal teams are still catching up.
• 30:31
  SPEAKER_01: So it's really important, more important than ever, that you be affirmatively reaching out to your legal team to get up to speed on this.
• 30:40
  SPEAKER_01: Be part of the AI policy drafting at your company, be part of those conversations because they don't know how you're using AI.
• 30:48
  SPEAKER_01: Why would they know?
• 30:49
  SPEAKER_01: They're not marketers or lawyers.
• 30:51
  SPEAKER_01: So you have to be part of the conversation.
• 30:53
  SPEAKER_01: And then the crisis plan, I made this um AI incident crisis response plan because people don't know the let me think about this.
• 31:04
  SPEAKER_01: I want to explain it, but I didn't know how many bad things could happen to a chicken until I talked to somebody that raises chickens, and I was like, oh my god, that sounds awful.
• 31:16
  SPEAKER_01: They can get sick, they get you know attacked by all different kinds of things, they get all kinds of weird ailments, and I was like, holy cow, I had no idea.
• 31:24
  SPEAKER_01: And this is exactly like that.
• 31:26
  SPEAKER_01: You have no idea how many weird bad things can happen to you as a result of AI if you have not experienced them before, which bless, I'm glad that you haven't experienced them before.
• 31:35
  SPEAKER_01: But until you've seen like a deep fake of your CEO go out on social, like doing some horrible thing, you don't know how to respond to that.
• 31:44
  SPEAKER_01: Okay, so this has a couple roles to play.
• 31:47
  SPEAKER_01: One of them is you think about the kinds of things that can happen to you and your team and you plan against them.
• 31:53
  SPEAKER_01: So this is like the worst possible thing that could happen with AI.
• 31:57
  SPEAKER_01: Let's sit down, pretend that it has happened, and figure out who's going to be impacted and what we need to do now.
• 32:03
  SPEAKER_01: So I've got this whole plan and just use it as a sample.
• 32:06
  SPEAKER_01: Don't like take this and just use it as your own.
• 32:08
  SPEAKER_01: Take it to your legal team, take it to your marketers, take it to your executive team, figure out if it works for you.
• 32:14
  SPEAKER_01: But down to running tabletop drills of something, like pretend something has just happened in relation to AI.
• 32:22
  SPEAKER_01: You've got a crisis now.
• 32:23
  SPEAKER_01: What are you gonna do in the next 72 hours?
• 32:26
  SPEAKER_01: What's the first thing you're gonna do?
• 32:27
  SPEAKER_01: First thing is you got to get a statement out there saying, Yes, we have seen the deep fake, we are aware of it, we're investigating.
• 32:34
  SPEAKER_01: No, it is not our person, it's not really our CEO.
• 32:37
  SPEAKER_01: Okay, that's step one.
• 32:39
  SPEAKER_01: Now you take it from there and walk through the next few days.
• 32:42
  SPEAKER_01: So then, God forbid the thing actually happens, you'll be ready.
• 32:46
  SPEAKER_01: So it's really just it's prep and practice.
• 32:48
  SPEAKER_01: It's like, what do they call it?
• 32:49
  SPEAKER_01: Uh, pre-mortem.
• 32:51
  SPEAKER_01: So you assume the worst possible thing has happened and you work backwards from there.
• 32:56
  SPEAKER_00: Well, I mean, that was really the biggest issue with having such great free or you know, basically free video generating tools that were so good at what they did, is it made it so much more likely that things like that would happen.
• 33:14
  SPEAKER_00: And it's so difficult.
• 33:15
  SPEAKER_00: People do not understand that everything you see on the internet is not real.
• 33:20
  SPEAKER_00: I'm still seeing so many of these fake, heartwarming stories about adoptable animals, especially on Facebook, because Facebook no longer allows you to report something as having been made with AI.
• 33:34
  SPEAKER_00: So they can post an entire story that's a complete lie and a video that's completely AI generated.
• 33:40
  SPEAKER_00: And you get all these nice, nice, you know, ladies I went to school with and their moms commenting on, oh, that's so adorable.
• 33:48
  SPEAKER_00: And you know, oh, how can we help?
• 33:50
  SPEAKER_00: How can we help support this?
• 33:52
  SPEAKER_00: So then they get scammed out of money from somebody putting up these horrible videos, and it makes me so mad.
• 33:58
  SPEAKER_01: I know it's it's terrible, it really is.
• 34:01
  SPEAKER_01: And education is the only solution for things like that.
• 34:04
  SPEAKER_01: So at work, you've got to educate folks about what your AI policy is, what tools they are and are not allowed to use, what they're allowed to use them for.
• 34:14
  SPEAKER_01: And I'd start with your hot kind of, hey puppy.
• 34:17
  SPEAKER_01: I'd start with kind of your highest risk, highest risk uses.
• 34:21
  SPEAKER_01: Can't speak.
• 34:22
  SPEAKER_01: Uh, like, for example, if you do mortgage brokering and you're gonna use AI to help you decide who does and does not get a mortgage, that is that is a very high-risk activity to engage in.
• 34:34
  SPEAKER_01: So you're gonna want to look at the EU AI Act, you're gonna want to look at whatever state or um or country in which you're operating primarily.
• 34:40
  SPEAKER_01: Make sure that you're not doing anything, like maybe you can't use it for that full stop, and you'd better figure that out, right?
• 34:48
  SPEAKER_01: Or if you're using it for marketing, are you using it to review your stuff?
• 34:51
  SPEAKER_01: Are you using it to help you plan and brainstorm, or are you using it to create stuff out of nothing?
• 34:56
  SPEAKER_01: Because that's a very that's like a higher risk use, uh, less less risky than hiring or firing based on AI decisions, but still risky.
• 35:05
  SPEAKER_01: So think about it like that from the beginning, and you'll be in a much better situation than someone that's going, I don't know, just use AI because you know the board wants us to use AI.
• 35:16
  SPEAKER_01: I mean, it happens, it's all over the place.
• 35:18
  SPEAKER_01: The pressure.
• 35:20
  SPEAKER_00: It was like how everybody had digital, you know, digital transformation was one of their top, you know, goals or top objectives, yeah, you know, for years running.
• 35:30
  SPEAKER_00: It was just everybody was doing it.
• 35:32
  SPEAKER_00: It was what everyone was talking about.
• 35:34
  SPEAKER_00: You couldn't not do it.
• 35:36
  SPEAKER_00: But, you know, I think we also we always see, because I know I would run into this with some of the guardrails, I would want to put in place from a um content governance standpoint and from just being a very um project organized, you know, organized kind of person who wanted to know how everything worked.
• 35:56
  SPEAKER_00: It's the you know, I went to school for journalism for magazine and I wanted to like run a magazine.
• 36:02
  SPEAKER_00: So, of course, I think I think in very specific compartments and processes and systems.
• 36:08
  SPEAKER_00: And a lot of people are like, hey, you're stepping on my creativity.
• 36:13
  SPEAKER_00: How am I supposed to work within confines?
• 36:16
  SPEAKER_00: Um, so I think that can be a big struggle too for a lot of marketing teams if you have folks who see themselves as creative talent for and you know, they're like, I don't want to have to talk to legal and compliance, and the more of these, you know, rules you shove at me, you know, how can I do my full creativity?
• 36:34
  SPEAKER_00: And you know, and they see people who made a killing off of creating, oh, what were they?
• 36:41
  SPEAKER_00: The um all the NFTs, like people who are creating art and selling those individual NFTs.
• 36:49
  SPEAKER_01: I hear you.
• 36:49
  SPEAKER_01: I wish I had talent and could have made an NFT.
• 36:53
  SPEAKER_01: There are plenty of people who made a lot of money, but that's not really about the art, is it?
• 36:58
  SPEAKER_01: Let's be honest.
• 36:59
  SPEAKER_01: Those NFTs, some of them are garbage.
• 37:00
  SPEAKER_01: It's like it's the risk premium off any new technology technology.
• 37:05
  SPEAKER_01: There's a whole bunch of folks like in Silicon Valley and elsewhere that have access to the very earliest of early days, the bloodiest bleeding edge stuff.
• 37:14
  SPEAKER_01: They'll get out there, and if they convince even like a handful of very early adopters to try it, that starts the wave.
• 37:22
  SPEAKER_01: Then they can skim the risk premium off the top.
• 37:24
  SPEAKER_01: And then, you know, if NFTs turn out to be worthless, which a lot of them are, and most of them are, they they don't care because they're already on to the next thing.
• 37:32
  SPEAKER_01: So that's what you have to, that's what you have to view everything through, is that there are people who will make money off of anything.
• 37:39
  SPEAKER_01: There are people who will broker risk and kind of pass it on as best they can to somebody else.
• 37:45
  SPEAKER_01: You can't be that person.
• 37:47
  SPEAKER_01: You're here for the long haul.
• 37:49
  SPEAKER_01: If you really are creative and you want to continue creating, you can't afford to get mixed up in stuff like this.
• 37:54
  SPEAKER_01: You can't afford to get mired down in a lawsuit because you didn't pay attention to AI.
• 38:00
  SPEAKER_01: Like take something of yours, put it into an AI, and have it help you add wind effects or something.
• 38:05
  SPEAKER_01: It's fine, but don't have it make up stuff for you wholesale.
• 38:09
  SPEAKER_01: Maybe it could help you animate something of yours if you put the sprites in, um, but don't have it create something brand new for you.
• 38:16
  SPEAKER_01: And why would you?
• 38:17
  SPEAKER_01: You're creative.
• 38:18
  SPEAKER_01: I just I definitely think you create because you need to create, not just because you're paid to create.
• 38:25
  SPEAKER_01: So why would you ever give that up?
• 38:26
  SPEAKER_01: But as far as compliance and like why would you want to be bothered with compliance?
• 38:31
  SPEAKER_01: You don't necessarily have to be the person that does it.
• 38:34
  SPEAKER_01: I mean, we don't send our designers off to talk to the lawyers, we just set up the system in which they operate and tell them the tools and give them the rules.
• 38:43
  SPEAKER_01: And the rules are not meant to constrict their creativity, the rules are just meant to prevent them from uh kind of exposing themselves to liability.
• 38:52
  SPEAKER_01: Yeah, because if you bankrupt the company you're working for, you know, oh no, you have no job now.
• 38:57
  SPEAKER_01: So it's it benefits everybody, but creators most of all, because the problem with using AI to create things is it's just really it's just taking it by definition, it can't be creative, it's taking stuff it trained on and representing it.
• 39:12
  SPEAKER_01: So unless they'd want that done with their own work, I mean, I think every creative person would recognize that's not something they want to do to someone else either.
• 39:20
  SPEAKER_01: So it's at its heart, if you go down to the heart of compliance, it's not just avoiding risk and money for your company, it's about doing the right thing by artists.
• 39:29
  SPEAKER_01: And I think every artist would appreciate that.
• 39:31
  SPEAKER_01: And by artist, I mean every creative person.
• 39:33
  SPEAKER_01: So writer, painter, graphic designer, cartoonist, I don't know, what other crazy sculptor, you know, creative.
• 39:41
  SPEAKER_00: And it can be it can be hard because you get you feel like, oh, but everybody's doing that meme where they create, you know, themselves as a Marvel character, or everybody's, you know, creating that meme where they're a Funko pop or whatever.
• 39:56
  SPEAKER_00: But literally, once you're saying the everyone's doing the thing in a brand name next to it, when you're talking about AI content, that right there is a red flag.
• 40:07
  SPEAKER_00: That means that you are infringing.
• 40:09
  SPEAKER_00: So it was if like there was the situation, I think Barbie did have a tool where you could create a Barbie version of yourself.
• 40:17
  SPEAKER_00: Well, as long as you used the Mattel tool to create your Barbie version of yourself, you weren't breaking their copyright because it was their tool that they provided for you to use for that.
• 40:28
  SPEAKER_00: But if you then went to a different platform and did that same thing, you know, then you are breaking the copyright and you could be infringing.
• 40:37
  SPEAKER_01: Yeah, potentially for sure.
• 40:39
  SPEAKER_01: So it's um it's interesting.
• 40:42
  SPEAKER_01: I don't know if I don't know if it would be worth it to anybody to do the same thing.
• 40:49
  SPEAKER_01: Like if if everybody's doing a thing, where's the differentiation in that for you as a marketer to begin with?
• 40:55
  SPEAKER_01: You should be doing something that people look at and go, oh, that's Erica's for sure.
• 40:59
  SPEAKER_01: Like she's the one who the she's got that snarky sense of humor, but like just on the line or whatever.
• 41:05
  SPEAKER_01: Like that's that's what you're going for.
• 41:07
  SPEAKER_01: You want people to recognize your stuff.
• 41:08
  SPEAKER_01: Like Chris Brogan's stuff, I would be really hard pressed to confuse his writing with somebody else's writing or Ann Hanley's.
• 41:15
  SPEAKER_01: Like you look at their stuff and you know that it's theirs, even if you never scroll to the bottom and read their their signature in the email.
• 41:22
  SPEAKER_01: So you don't want to be doing anything everyone else is doing, but also when it comes to things like trademark, like copyright, trademark's an easy one.
• 41:30
  SPEAKER_01: You're not ever going to be selling something that looks so like somebody else's product that it could be confusing.
• 41:35
  SPEAKER_01: So if like a moron in a yeah, so like if a moron in a hurry, that's the standard.
• 41:39
  SPEAKER_01: If a moron in a hurry might make a mistake and buy yours thinking it's theirs, then you have a problem from a trademark standpoint.
• 41:46
  SPEAKER_01: Okay.
• 41:46
  SPEAKER_01: But for copyright, that's a whole other ball of wax because now you're in this, like, it's a very difficult uh thing to prove that you did not infringe if it looks exactly like somebody else's thing.
• 41:59
  SPEAKER_01: So even if you're like, I never even heard of Erica before, I sure didn't read her stuff.
• 42:03
  SPEAKER_01: Well, why does your thing sound exactly like her?
• 42:06
  SPEAKER_01: Why does it use a couple phrases that she's used that are really distinctive, you know, and why does it make jokes that Erica has made before?
• 42:13
  SPEAKER_01: And you're like, Why does it talk about her cat?
• 42:15
  SPEAKER_01: Yeah, you're like, wait a second, I actually did possibly use AI.
• 42:19
  SPEAKER_01: Maybe it, you know, when I asked for snarky humor, maybe that's where it shopped it around from, you know.
• 42:24
  SPEAKER_01: So you you've got to be thinking about these kinds of things.
• 42:28
  SPEAKER_00: So I'm hoping um that you have some ideas for, you know, some kind of like easy-to-remember tips for folks, how they can make sure that they're being safe when they're using AI.
• 42:40
  SPEAKER_00: I think we've talked about some of them.
• 42:42
  SPEAKER_00: Um, you know, one things that you talked about, obviously, was you know, just making sure that you take the time to tell AI what not to do.
• 42:49
  SPEAKER_00: Don't assume it has the same moral compass as you.
• 42:52
  SPEAKER_00: It has no moral compass.
• 42:53
  SPEAKER_01: No moral compass.
• 42:54
  SPEAKER_00: Yeah.
• 42:54
  SPEAKER_01: No moral compass.
• 42:56
  SPEAKER_01: No, you're on to that.
• 42:57
  SPEAKER_01: Um, part of it is just how you introduce it in the first place into your place of business matters a lot.
• 43:04
  SPEAKER_01: So the correct answer is not to just start doing something, anything, because you have to start using AI.
• 43:11
  SPEAKER_01: It's like, let's look at what we've got, let's look at who we've got, let's figure out where it makes sense to deploy it for manual work, non-risk, you know, like less risky stuff.
• 43:21
  SPEAKER_01: Um, let's think about it and then do it in an intentional way.
• 43:25
  SPEAKER_01: Have an AI policy and train everybody on it.
• 43:29
  SPEAKER_01: Have an AI crisis plan and practice it.
• 43:32
  SPEAKER_01: Don't just have it and file it, have it and practice it.
• 43:35
  SPEAKER_01: Do the tabletop drills.
• 43:36
  SPEAKER_01: Then when you get these tools that everyone's saying that they want, before you deploy them org wide, configure them a little bit.
• 43:44
  SPEAKER_01: So, some connectors you're gonna want to enable, some connectors you're not gonna want to enable.
• 43:49
  SPEAKER_01: You have to decide which is which.
• 43:51
  SPEAKER_01: Don't let it go down to the people and be like, you know, God sort them out.
• 43:55
  SPEAKER_01: No, make your plans in advance.
• 43:57
  SPEAKER_01: If you don't want people messing around with certain And proprietary systems, connecting them up to your AI, putting that data at risk, don't allow the connection.
• 44:06
  SPEAKER_01: So a lot of it is just preliminary planning.
• 44:08
  SPEAKER_01: And it's not like too late.
• 44:10
  SPEAKER_01: If you've already got people cropping up across the org using AI for different things, they're going to talk about it because they want people to know that they're using AI and that it's the new shiny thing.
• 44:20
  SPEAKER_01: They're raising their hand.
• 44:21
  SPEAKER_01: Now I'm not suggesting you go come down on them.
• 44:24
  SPEAKER_01: I'm suggesting you go, hey, this is pretty cool.
• 44:26
  SPEAKER_01: Um I'm a little concerned that it seems to be connecting with like our customer database.
• 44:31
  SPEAKER_01: I feel like maybe let's not do that.
• 44:33
  SPEAKER_01: But everything else is great.
• 44:35
  SPEAKER_01: Like maybe use personas instead of using actual customer data in your thing.
• 44:38
  SPEAKER_01: Pretty pleased before we get sued.
• 44:41
  SPEAKER_01: Thanks.
• 44:42
  SPEAKER_00: Yeah.
• 44:42
  SPEAKER_01: So yeah.
• 44:43
  SPEAKER_00: Well, I mean, it feels like we can learn a lot from you know social media's proliferation, you know, when it was still a new thing 15, 16 years ago.
• 44:52
  SPEAKER_00: And, you know, frequently it would be an individual within an organization would go out and create, you know, accounts with the company's name and go, hey, I'm doing this thing because I love our company and you know, and I'm in marketing or I'm in customer service.
• 45:09
  SPEAKER_00: And, you know, I felt like we needed to have an account.
• 45:12
  SPEAKER_00: I know one of my good friends was the one who had actually grabbed the Twitter account for our employer back in the day because she was like, I can't believe we didn't grab our company name yet.
• 45:23
  SPEAKER_00: So she grabbed it, which was amazing because it meant that we didn't end up having somebody who hated the brand grabbing it.
• 45:29
  SPEAKER_00: But on the flip side, then in you know, some other countr countries, there were situations where local, you know, local offices created social accounts.
• 45:40
  SPEAKER_00: So I mean, I do feel like we probably can learn a lot from the proliferation and and how companies kind of got better governance and uh workflows in place around using social.
• 45:52
  SPEAKER_00: And you can probably apply a lot of that to AI.
• 45:55
  SPEAKER_01: The fundamental rules apply, yes, to this and everything else.
• 46:01
  SPEAKER_01: So just like with social media, you could not pretend to be someone else.
• 46:05
  SPEAKER_01: You cannot pretend to be someone else using AI.
• 46:08
  SPEAKER_01: You know, don't create fake videos of Howard Stern promoting your brand, you're gonna have a bad time.
• 46:14
  SPEAKER_01: Not even if people believe it's him, you're gonna have a bad time, right?
• 46:18
  SPEAKER_01: Either way.
• 46:19
  SPEAKER_01: Um, the same with the work that you're producing.
• 46:21
  SPEAKER_01: You can't use AI to produce work when you have no idea where it came from.
• 46:25
  SPEAKER_01: You need to document the pieces of it that were human-generated.
• 46:29
  SPEAKER_01: It gives you a better chance at protecting copyright in those things.
• 46:32
  SPEAKER_01: And if you don't care about protecting copyright so much, if you sell widgets and you're like, don't care, I don't care if they take my blog post, that's fine.
• 46:40
  SPEAKER_01: Um, that's okay to a point, but you still don't want to be on the hook for copyright infringement, right?
• 46:45
  SPEAKER_01: So you'd still better be really careful about what you're doing.
• 46:48
  SPEAKER_00: You can't just point to and go, oh, it was just the the AI agent did it, you know, not our fault.
• 46:54
  SPEAKER_01: No.
• 46:54
  SPEAKER_01: Well, this is the other thing.
• 46:55
  SPEAKER_01: So this brings me to the question of AI vendors.
• 46:58
  SPEAKER_01: You want to be really careful which AI vendors you work with.
• 47:01
  SPEAKER_01: Pick somebody responsible.
• 47:03
  unknown: Yeah.
• 47:03
  SPEAKER_01: And ask them where does it, where, you know, where are your models, what are they trained on?
• 47:07
  SPEAKER_01: Where do your models come from?
• 47:08
  SPEAKER_01: Where do they get their data?
• 47:09
  SPEAKER_01: And if they're like, honey, don't worry about it.
• 47:12
  SPEAKER_01: There's a couple of reasons to run the other way.
• 47:14
  SPEAKER_01: Number one, they called honey.
• 47:16
  SPEAKER_01: Number two, they said don't worry about it.
• 47:18
  SPEAKER_01: You want to know up front where that stuff's coming from.
• 47:21
  SPEAKER_01: There have been companies that started and their whole business model was like scraping other people's content and presenting it in a new way.
• 47:30
  SPEAKER_01: Absolutely not.
• 47:31
  SPEAKER_01: Not for them, not for you.
• 47:33
  SPEAKER_01: You don't want to like buy into that type of liability because then you're paying for the privilege of getting sued.
• 47:38
  SPEAKER_01: So, no thanks.
• 47:39
  SPEAKER_01: So ask your vendors the hard questions.
• 47:41
  SPEAKER_01: If they won't answer them, run away because you don't need that in your life.
• 47:45
  SPEAKER_01: Pick someone who's gonna lay it on the table and show you how the sausage is made.
• 47:49
  SPEAKER_00: 100%.
• 47:50
  SPEAKER_00: And I think that's you know, that's great advice for many things in life.
• 47:54
  SPEAKER_00: Um and I think it's a good place for us to probably stop for today.
• 47:59
  SPEAKER_00: But I want to know, Carrie, where can people find more from you about this and other marketing-related things?
• 48:07
  SPEAKER_01: Well, I'm on LinkedIn, and when something has got me concerned, I will post about it on LinkedIn.
• 48:14
  SPEAKER_01: Um, you can find more information about me at carrie.show or carrygorgone.com.
• 48:19
  SPEAKER_01: They go to the same place.
• 48:21
  SPEAKER_01: I'm still I'm like, I have my Twitter just to hold the account, but I don't post there anymore.
• 48:26
  SPEAKER_01: So LinkedIn is the best.
• 48:27
  SPEAKER_00: I had held mine for a very long time, but I ended up nuking all of my accounts there, and it felt bad, but I eventually did it.
• 48:35
  SPEAKER_01: I don't I wouldn't feel bad.
• 48:36
  SPEAKER_01: I still want someone else to like have my thing.
• 48:39
  SPEAKER_00: Oh, I felt bad because I loved Twitter so much, it was such a great place.
• 48:44
  SPEAKER_00: Like 2010 Twitter, yes, exactly.
• 48:47
  SPEAKER_00: Um, but for me, it was once they changed it so that people that you had blocked could now see all of your content and you know, reshare your content and all of that kind of good stuff.
• 49:01
  SPEAKER_00: I was like, well, that doesn't really work so great for me anymore.
• 49:05
  SPEAKER_00: So I actually just deleted everything, just deleted everything.
• 49:09
  SPEAKER_00: I'm like, no, thank you.
• 49:11
  SPEAKER_01: Um so Carmen said, Bummed, I got here late.
• 49:15
  SPEAKER_01: That's okay, Carmen.
• 49:16
  SPEAKER_00: Carmen, you can watch the replay any minute.
• 49:18
  SPEAKER_01: Just replay, or you can just find me on LinkedIn and I will I won't give people legal advice because I don't practice anymore, but I will give you information.
• 49:26
  SPEAKER_01: I'm happy to share information and uh I'll make a face because I can't control my face.
• 49:32
  SPEAKER_01: So if it looks concerned, that tells you something.
• 49:36
  SPEAKER_00: I love it.
• 49:37
  SPEAKER_00: Well, I hope that I'll get to see you in person at some point later this year.
• 49:41
  SPEAKER_00: Um, but until then, I'm gonna go and look at those super cool Google Docs that you shared with me and that.
• 49:50
  SPEAKER_00: I'm gonna share in the uh show notes and the uh and in the recap.
• 49:56
  SPEAKER_00: And yes, so I do run basically a zoo here in my house.
• 50:00
  SPEAKER_00: This is Ash, and he is five months old.
• 50:05
  SPEAKER_01: Look at that precious baby.
• 50:07
  SPEAKER_01: I know I'm waving like a cat carries.
• 50:09
  SPEAKER_00: He can see you.
• 50:11
  SPEAKER_00: He looks, he's like, Oh, who's the lady waving at me?
• 50:15
  SPEAKER_01: Oh, thanks for calling me a lady cat.
• 50:17
  SPEAKER_00: That's why that's why it's fun to do these live streams on video and not just have an audio only podcast.
• 50:25
  SPEAKER_01: Oh, I agree.
• 50:26
  SPEAKER_00: So if you're just listening to the audio only, you missed a heckin cute dog and a really cute cat, too.
• 50:32
  SPEAKER_01: So that's true.
• 50:34
  SPEAKER_01: Those will be the spikes on YouTube.
• 50:37
  SPEAKER_00: That's right.
• 50:39
  SPEAKER_01: What parts do people watch?
• 50:41
  SPEAKER_00: I'll make sure to uh put chapters in for both of those.
• 50:45
  SPEAKER_01: Thank you.
• 50:46
  SPEAKER_00: Well, Carrie, I love you to pieces, and thank you so much for being on the show today.
• 50:51
  SPEAKER_00: And everyone else will catch you soon on the next edition of Content Chat.
• 50:55
  SPEAKER_00: Bye, everyone.