AI Hope

Europe just moved on AI slop. The new rule forces the big AI companies and the startups riding their coattails to watermark whatever they generate, images and text alike, in a way ordinary users like you and me cannot strip back out. Anthropic was the first major player to implement mandatory watermarking, and given Europe's market size, the rest of the industry will fall in line whether they are enthusiastic about it or not.

I think the effect lands unevenly, and mostly in a good way. Writers and artists who create without leaning on these tools stand to benefit, because watermarking makes their work newly distinguishable in a feed that has been drowning in AI output. 

Companies with a legitimate reason to use AI, a small organization that could never otherwise afford a training video vendor, for instance, get more transparency without more shame attached to using it. Where the rule will probably do the least is in everyday content consumption. 

Platforms that already profit from AI generation, and Meta is the example I keep coming back to, have little incentive to throttle their own output just because it carries a label, and most people scrolling past a watermark are about as likely to read it as they are the terms of service on a credit card statement.

The disclosure opens up a real artisan's market. An ad agency or a studio can now say a piece of work is human made and mean something specific by it, the same way a Bourne movie's practical stunts read differently than overwrought CGI-ridden action adventures. 

Music might be the one category where people keep caring regardless of what the market does elsewhere, since a computer-generated jingle is a much harder sell than a human one. I do not think that changes.

I do not expect watermarking to end the slop problem. Bad actors have had a decade-long head start on every enforcement mechanism built to catch them, and that head start does not close because of one regulation. But I still call this a hopeful moment, because it is an industry being pushed into a transparency it would not have chosen for itself, and a market finally starting to reward the people who are still doing the work by hand.


00:00:13 The EU's new AI watermarking regulation 
00:01:05 Anthropic moves first, the rest of the industry will follow 
00:01:50 How social networks will use watermarks (and why Meta won't) 
00:03:20 Why bad actors will always stay ahead 00:03:48 Where AI-generated content is genuinely appropriate 
00:04:47 The credit card statement problem: nobody reads disclosures 
00:05:28 The rise of an "artisan's market" for human-made work 
00:06:57 Music: the one place people still care 
00:07:29 How each AI vendor will likely respond 

What is AI Hope?

What's your AI success story? What made you realize you had to change? And what does a hopeful future with AI look like to you? AI Hope with Geoff Livingston asks three simple questions of people doing the hard work of adapting and thriving in the AI era.

Intro
---

Geoff: Hey everybody. My name is Jeff Livingston, and this is AI Hope. My book, Now Is Gone, is about never-ending change. This podcast is about what comes after, the stories, the people, and the reason for hope.

EU AI Watermarking Regulation
---

Geoff: Hello, friends. Hope you are doing well. This is AI Hope, episode number 10, and today we don't have a guest, so we're gonna talk about a trend, and that trend that I'd really like to talk about is watermarking. So specifically what's happened is the EU has approved kind of a regulation, if you would, to ~kind of, uh, ~ratchet back some of the slop that's being created and make people disclose when they're using AI.

And basically how they're doing that is forcing the big tech companies and the AI startups to watermark their various generations if you would, generations of images and text, and to, ~uh,~ make that something that people, users like you and [00:01:00] me, cannot remove. ~Um, ~that's ~kind of ~interesting. The first major player to do that was Anthropic.

~Uh, ~obviously, the rest of the industry is gonna follow suit. ~Uh, ~Europe's just way too big of a market for the tech industry to lose. So what does that mean, right? What-- How is that valuable? I know a lot of people are upset by that in the AI side, and I think a lot of people that are normal human beings are excited by it.

And I think that what's gonna happen is a few things. One, I think it's actually gonna be really good for AI. I think it's gonna be really good for people that are actually good writers and good generative, ~uh,~ creators on their own without these tools, so basically artists, right? And I think it's actually not gonna mean much for a lot of people that are specifically consuming content.

Social Networks & AI Slop
---

Geoff: A-and here's how that plays out. So in my mind, the first thing that happens is obviously some social networks are gonna use that as a flag, [00:02:00] right? They'll be able to see that watermark and not source that content. Does that mean every social network will take that initiative and stop sourcing slop? ~Well, ~I think Meta probably won't because they're a creator of AI tools, and they wanna encourage people to use it.

In fact, what might happen in the case of Meta, in particular on Instagram and Facebook, and if anybody actually still uses Threads, ~um,~ I know people do, I'm just being smart. What might happen there is that meta AI-created content surpasses other created content. ~Um, ~I know YouTube is already flagging this stuff.

They're asking users to, ~uh,~ basically identify... I know LinkedIn is asking users to basically identify slop. I expect some of the other social networks will use this as a means to basically throttle down [00:03:00] slop. ~Uh, ~so some people won't see as much content, some networks maybe you will. And one thing I've definitely learned over the past 10 years in technology, even though I've been in the business for 30 years, but particularly my time in the cybersecurity side of the space, bad actors are always ahead.

AI Content in Business
---

Geoff: So I don't necessarily think that we'll get rid of the problem completely, and in some cases, it's just gonna spur innovation from some AI-oriented creators that are gonna go ahead. So that's one with social networks. But two, I think companies that are creating AI content, ~uh,~ are going to do so on their websites and in their various materials to their heart's content and not care about it, and maybe they shouldn't.

~Um, ~and here's why, because in some cases, and I hate saying it, I know this will anger some people, but AI-generated content is appropriate. It is appropriate for things [00:04:00] like training videos, for example, where the company cannot afford to hire ~a, ~a different vendor to create the training videos. And using tools like Synthesia and other, ~uh,~ various different, ~uh,~ content providers and AI providers, they are able to do that.

~Um, ~I just think that what it does do is, ~uh,~ it just creates that level ~of, ~of transparency that if somebody wants to check to see if something is AI-created, they can. And in the case of things like RFPs, that may make a difference for some companies, and for many it, it may not. ~Um, ~I think what we are gonna unfortunately discover is how lazy we as human beings can be about checking things.

So the classic example of this is the credit card statement, the rules or the terms of service for using the credit card. Everybody gets outraged when they read them, and then [00:05:00] they go ahead and use them and never think twice about them. ~Uh, ~and in many cases, some people just don't ever look at them, ever.

And I think we're gonna see that with AI. ~You know, ~people are not gonna care what the content is. In fact, I think a lot of the general public just, even though there's a, a lot of concerns and a lot of negativity right now, people are still consuming this. I see it all the time in the feeds, and I see it gets liked so much more than regular content.

The Artisan Market
---

Geoff: Slop is out there, and in some cases it's winning, and I think some people just don't care. All right. ~Um, ~so how does that help content creators, right? ~Well, ~I do think that there is going to be a higher-end market for well-created content. In some cases, that'll be hybrid, hybridized, where a very strong creative uses these tools to help build some of their content But also it's going to create kind of what [00:06:00] I, I would call an artisan's market, where an ad agency can say, "This was human-generated."

Or a movie company can go back to, ~uh,~ scene pieces and say, "We are not using AI. We are not even using CGI. We are using set pieces. We are creating a true Hollywood movie for you." ~Um, ~and there may be a market for that. I bet there is actually in some cases. ~Um, ~and I think if you look at Jason Bourne, for example, as a movie series, one of the things that was increasingly refreshing about that movie series was the lack of special effects.

It was just hard, well-cut movies, and it got overplayed, obviously, that style of creating content, but I think that's why that was so exhilarating and interesting as an action series, so much more different than what we're used to seeing at that time. So I do think for artisans, ~um,~ any different field within the creative space, this means a lot.

~Uh, ~

Music & Vendor Stance
---

Geoff: I think the one area where it means everything is [00:07:00] music, where people are getting very upset by AI-created music and, ~uh,~ I think with good reason, right? I don't necessarily wanna hear a computer-generated jingle either. I wanna hear somebody's artistic expression. And then, ~uh,~ the last place where it helps is AI, and that allows the vendors to say, "Hey, man, we are behind this regulation, and we support disclosure, and we will not allow anybody to create content without these watermarks."

I think Anthropic leading the way is really smart for them. As usual, when compared to ChatGPT and OpenAI, they're ahead. ~Um, ~I think with Google, ~uh,~ it's probably in their best interest. We all know that all their, ~uh,~ search little, ~um,~ summaries, if you would, are created with AI and generative tools like Gemini.

But, ~uh, you know, ~Google is a little more in the middle ground there, and like Meta, they have a vested interest in using AI to their, ~uh, uh, ~to their own value proposition. [00:08:00] But they already disclose, so I don't think it's a huge hit for them. So again, right down the line, vendor by vendor, these guys are going to have to kowtow.

I do think the way that they do it and the way they make their watermarks available to, ~uh,~ the general public and how they are disclosed or found is gonna make a big difference, but I do think it will help them, and I also think it'll help AI content be more accepted in the long term because it's clear, it's transparent, and again, I think a lot of people just don't care.

I think they just wanna know what it is, and fine, I know what I'm dealing with, and I will look at it through that lens, and regardless, I need this information, and I want to consume it.

Conclusion
---

Geoff: So that's my thoughts on AI watermarking. I'd love to hear yours, and I do think it's actually a good, hopeful moment for the industry, and I'm happy to talk about it today.

So with that, that is your weekly AI Hope. ~And with that, I am out of here. Cheers. Oh, I forgot.~ Please punch the [00:09:00] like button. Please subscribe. Please leave a review for this podcast. I appreciate your time. Cheers