Is It OK for Platforms to Train Their AI With Your Content?

When is your content, our content?

In the movie Fast Times at Ridgemont High, Jeff Spicoli (Sean Penn) has a pizza delivered to his history class taught by Mr. Hand (Ray Walston), who isn’t pleased.

[youtube https://www.youtube.com/watch?v=E81eW3BFLHs?feature=oembed&w=500&h=281]

“Mr. Spicoli, you’re on dangerous ground here. You’re causing a major disturbance on my time.”

“I’ve been thinking about this, Mr. Hand. If I’m here, and you’re here, doesn’t that make this OUR time?”

“You’re absolutely right, Mr. Spicoli. It is OUR time.” Then, Mr. Hand invites the entire class to eat the pizza and takes a slice for himself.

It’s a perfect metaphor for where big technology seems to be headed with your content.

Marketers think, “If my content is here, and your content is here, doesn’t that make this OUR platform?”

And the tech companies say, “You’re absolutely right, Marketer. It is OUR platform.” Then, they proceed to take your content and feed it to their learning models for their generative AI tools.     

In the past week, Adobe Creative Cloud customers grew outraged over a new policy that seemed to give it access and rights to any users’ content. Meta, the parent company of Facebook and Instagram, informed the U.K. and other European users that its new privacy policy allows their information to be used to develop and improve its AI products.

What should marketers make of all this? Should you sit back and watch your content pizza get distributed to the class? Do you even care about sharing it with the class if that’s what is necessary for it to become the best answer given by generative AI?

We asked Robert Rose, CMI’s chief strategy advisor, to answer the tough questions. Watch or read on for his take:

[youtube https://www.youtube.com/watch?v=UxD0wgctdvc?feature=oembed&w=500&h=281]

Social media has long benefited from your content

A long-learned lesson in social media is that your content helps build their audiences.

In the halcyon days of social media, marketers bet on organic reach. You could use social media to reach an aggregated audience without spending advertising dollars. You just needed to create engaging, wonderful, differentiated content that resonated with audiences on whatever platform you used.

Of course, as these things go, every marketer in every industry got that same memo and soon organic reach began to fall. So, you started paying for the privilege of making sure your content got in front of those audiences, and social media became just another media channel.

But you didn’t give up hope. You didn’t stop producing content for that organic reach and engagement. You put some of your most valuable thoughts in that content, thinking if it’s really good and really engages the audience, you would build enough interest that whatever organic reach it would get would be valuable to your marketing.

We could have a great debate about whether those days are still here, if they’re long gone, or if they ever really existed. But what is evident is that these platforms are now using your content in new ways to build value for themselves.

Do you even have a choice in the matter?

In the last month, Meta surprised many Facebook and Instagram users by revealing that content on the platforms trains the company’s AI image generator. Hundreds of thousands of Instagram users reshared a message on the platform telling the company they did not consent to using their data to train AI.

But that declaration is worth as little as that meme from your mom telling you to cut and paste the denial to use your information on Facebook. You signed away all those rights when you started using the platform. 

The truth is more complicated. Technically, you or your employer owns a copyright on the original content you post on Facebook or Instagram. But the terms of service your company originally agreed to granted Meta licensing rights to use that work as it sees fit. It can duplicate, distribute, learn from, and modify it. In most cases, it can create derivative works from it. (Think about those year-in-review or memory posts it creates.)

In fact, get ready for a tasty slice of irony. Meta usually has more license to use your company’s imagery and content than you do as an employee.

Let that sink in a little bit.

Drama seems about a lot of nothing

But now, these platforms are going further. Just last week, Adobe found itself on the wrong end of the pitchforks and torches when it modified its terms of service. This phrasing most upset its users: “(W)e may access your content through both automated and manual methods, such as for content review … and using techniques such as machine learning in order to improve our Services and Software and the user experience.”

Well, that didn’t work out great for Adobe.

Social media users (irony duly noted), tired of their acquired role as epidemiologists and economists, decided they were now lawyers. They concluded that Adobe wanted to spy on customers’ content and use it to train its large language models.

Any sane review of the service term changes would likely think Adobe wanted to cover itself for using AI features to automate some services it provides to customers. To do that, it must know what the content is.

Anyway, I won’t pretend to be a lawyer, but this feels like a lot of drama about nothing. But the conversation does bring up a relevant point.

Should you be OK with your content on any platform – a proprietary one for private content development (e.g., Adobe Creative Suite, Microsoft Word, Google Docs) or a public one (e.g., Facebook, Instagram)?

The answer isn’t clear.

More than a handful of consulting clients have come to me in the last few months truly worried about allowing Google or Facebook to have their content. They fear their wonderful thought leadership content and inspiring customer stories will end up as fodder for the learning model and become the answer to a prompt on ChatGPT or Gemini.

That concern is legitimate.

They wonder if they should trust companies like Google, Microsoft, and Adobe that could access proprietary, private, and sometimes sensitive information to train LLMs.

Again, the concern is legitimate.

By the way, Apple just announced its new AI feature that will create actions based on everything you’ve done on your personal device, such as playing the podcast your wife mentioned in that email.

The only real conclusion for marketers

It all brings up the one lesson and only way for marketers to think about this.

There’s a big difference between the content you post publicly to reach and engage an audience and the content you keep private.

I can only see the benefit if your content signals AI to become the answer for a search or to become part of its derivative content for others. Creative marketing has always been a copycat business. Great campaigns, creative, copy, and thought leadership have always been unique and extraordinary until they aren’t.

As human marketers, you have always learned from and adapted your predecessors’ thought leadership, blatantly stolen your competitors’ copy, been inspired by someone else’s advertising copy or tag line, tweaked authors’ content to match your point of view and brand. The only thing AI does is speed up that process.

So, what do you do about platforms using your content to train their AI? Lean in. If you don’t want to play the game, stop using the toys. But, if you learn how to play the game, you can also win it.

As Spicoli might say, “That’s awesome. Totally awesome.”

Want more content marketing tips, insights, and examples? Subscribe to workday or weekly emails from CMI.

HANDPICKED RELATED CONTENT:

Cover image by Joseph Kalinowski/Content Marketing Institute

Leave a Reply

Your email address will not be published. Required fields are marked *