It's July 2023, and AI developments are still rocking the world every week. Voice cloning from a 3 second sample. Text to video. Full 3-D objects and indoor spaces AI-assembled from a handful of photographs.

The world is a crazy cool place, full of possibility.

But with that possibility comes problems- voice-cloning scams that are even harder to detect and more emotionally targeted. Risk regarding stock markets and AI trading. Concerns about military robots going rogue, or people falling in love with AI and we just stop making new people.

Every big tech change has comes at a cost. Electrocution wasn't very common before we "invented" electricity. Car accidents never happened before cars. Airplane crashes never happened before airplanes.

Big advancements come with big risks and big adjustments.

With AI, one of the immediate concerns has been around the protection of creative work, and the livelihood of artists - and most of this concern is about how AI challenges copyright by "plagiarizing" the hard work of artists everywhere.

Technology is a knife. Whether it's helpful or harmful depends on who wields it.

Let's look at some of the big problems we're confronted with on this issue.

Because whatever happens from here, AI is the biggest airplane of all, and we're all on it.

Buckle up. We've no idea who's flying this thing.

The big problems of AI and copyright protection

Technology is a knife. Whether it's helpful or harmful depends on who wields it.

Lawmakers are very bad at technology law

“Why can't copyright policymakers do something?”

When it comes to technology, lawmakers have an atrocious track record.

5 years ago, it was decided that Internet "cookies" might be a bad for you, and that you needed to agree to them. Unfortunately, the web is "stateless," which means that when you click to check out that cool new hoodie on a different page, the website has just forgotten who you are, and everything else you were buying too.

Can you imagine if in Uber Eats, you had to order, pay for, and deliver each item separately? Websites rely on cookies for everything from content analytics and marketing, to user preferences, login capability, and shopping carts.

The internet is pretty useless if every site has severe Alzheimer's.

The result was billions of dollars of development putting cookie consent banners on pretty much every website in the world. Now, every website you visit, you have to read a legal contract and click yes.

Is your life better now?

Now, I get it- and I'm sure the intentions of those policymakers was good.

But in my personal view those technical requirements should never have been enforced on website owners. It should have been discussed and implemented as part of the HTTP protocol standard, and become an integrated part of web browsers themselves.

You choose what settings you are comfortable with, and when you want to be notified for specific approvals. You decide what information can be captured, and stored and for how long- and it's all done smoothly, efficiently and consistently with minimal disruption to you. All with basically zero cost to the website owners.

Now that's a better Internet.

There's always a chance that some intelligent policymakers will assemble a solid team of tech geniuses and hammer out some AI policies that make sense.

But history is simply not on their side.

Creativity and originality are philosophy problems

“Isn't all AI-generated content a form of plagiarism?”

The question that sparked this article was framed on the idea that because AI is trained on existing works, that anything it creates must be at least partly based on those works.

That's probably mostly true.

But is our brain trained any differently?

We're all taught from the same stories and "classics." 99% of the ideas we have are handed to us by our parents, friends, and culture. In fact, that's part of the very definition of culture.

We're handed the same 26-letter English alphabet and 88-key piano. How much originality can there be? Everything we create is a mix of what we've seen, tasted, and experienced, plus a bit of randomness.

Just like AI.

Even in writing there are thought to be only seven basic plots [4] . Have you ever wondered why Star Wars and Harry Potter seem so incredibly similar?

And those were both created by humans.

“Good artists copy, great artists steal.”
- Pablo Picasso

Originality was always tricky, and it's very hard to draw clear lines here. If your voice sounds exactly like Paul McCartney's, are you allowed to sing in public?

An AI that sounds like Drake is considered infringement by some.

Before we can make laws that work, we have to revisit basic ideas like philosophy of ownership, our definitions of "creativity" and "originality", and why we're doing all of this anyway.

Better Copyright won't solve anything

Content creation and copyright are in a very weird place right now.

If you are watching AI and where it's going, you're probably aware that copyright is quickly becoming pointless. The reason is that even if we can precisely define "too similar", and accurately measure it to identify infringement... an AI can, too.

Give me any well-defined copyright law, and I can make an AI that generates 100% legal, non-infringing material that is just 1% outside of that line.

In April, a TikTok user by the name "Ghostwriter977" cloned Drake's voice and The Weeknd's musical stylings, and turned out a never-before-heard piece of music called Heart on my sleeve (ft The Weeknd).

Everyone hearing it thought that it was Drake, and The Weeknd. And most of them loved it, which totally pissed off Drake, and Universal Music Group (UMG). They immediately filed takedown notices, and the song disappeared from some sources while it was cloned and popped up on others.

It is the Internet, after all.

Deep Tom Cruise is another great example of where these lines get very... very blurry.

These are examples of deepfakes- content that was produced with the intention of being indistinguishable of the people or voices portrayed.

But even then- are they copyright infringement? We're still not sure. And even if we knew, we've no idea how to prevent it from happening again.

Copyright isn't the real issue, anyway

Let's fast forward a year or two to a point when AI can generate anything you like- and it knows how to ensure that it avoids any form of plagiarism or copyright infringement.

Are artists safe then?

I don't see how.

With the exception of artists who create truly exceptional pieces that absolutely capture the love and attention of the media and the public, very few will be able to produce writing, art, music that can be sold at an appreciable price.

We can already see that impact happening in related creative industries that demand slightly less uniqueness .

If you ask me which industries will be crushed first by AI, I can name two...

1. Stock photography companies. These will really have no purpose, as it takes me longer to find a photo than it does to just prompt it in Midjourney. And that's already the case today for 99% of the artwork I use.
2. Professional models. With the exception of very famous models who have a personality and brand established- why would you pay someone to sit for hours at a photo shoot, when you can just say "AI make me 50 model shots wearing this designer dress." Zero photoshopping needed, and the actual people don't even need to exist at all.

If the purpose of copyright is to protect a creative professional's livelihood, it can no longer provide that protection.

The whole World has to agree on the same rules

Has that ever happened? On anything?

The image I created for this article was created with Midjourney. I simply asked it to draw "The Internet." It decided that the Internet has a lot of cats. Seems legit.

I "made" this image. But who owns it?

According to current U.S. copyright law, machine generated content cannot be copyrighted. Except, that changed with the advent of cameras, and photoshop, and electric guitars, and the line moved a bit.

Now the determination is more about "creative artistic control". If you have a reasonable degree of control over that machine, and had substantial influence into what it produced, your work can be protected by copyright.

Cameras work because you choose the lenses, settings, and of course where and when to hit that button. Photoshop also works because you started with a blank page, and made a lot of artistic decisions in turning out that cool piece of art.

But with current Large Language Model (LLM)-based AI's... not to much.

I prompted Midjourney to draw "The Internet," and it dutifully spat out four images. They were all good but I liked this one best, and I've named my new masterpiece...

Hacker Jesus.

"Hacker Jesus and the 12 cats."
A totes-original artwork by Michael Wells ( and Midjourney )

According to U.S. law, I can not claim any copyright on Hacker Jesus, because it was created by an AI. In the view of the U.S. copyright office, there just wasn't enough creative control. I typed "the internet", and the Midjourney machine did the rest.

Therefore, according to U.S. law, I have no rights to this image. It is unprotectable and effectively in the public domain.

But Midjourney's terms of service say that because I created it, I have exclusive rights to it. No one else is permitted to use it.

So what rights does that confer? Am I only protected from other Midjourney users, who also accepted the same ToS? Who enforces my rights?

Confused? I've just gotten started.

I happen to live in New Zealand at the moment, and NZ copyright law is different. Anything I create with a machine - including AI - is mine. I don't even have to register it. Copyright is automatically assigned to me the moment I create it.

Cool, so... I own the copyright of Hacker Jesus in NZ, but not in the rest of the world?

Not so fast. The U.S. and NZ are both parties to the Berne Convention which means that the US must protect my NZ-based copyright.

It would take a multinational legal team to figure this out, and probably a U.S. Supreme Court case as well.

In the end, there's only one thing I know for certain...

We live in interesting times.