😸 Major AI lawsuit proceeds

Welcome, humans.

Well, we’ve come full circle: first, AI makes videos in the style of humans; now, humans are making videos in the style of AI.  

To prove how accurate this is, there’s actually been a lot of absurd AI videos we wanted to share in “What gives it away that this is AI” that were a little too obvious…

We’re talking about videos like this one, where motorcycles blast off over explosions, or this guy, who jumps in the air and then melts into the ground.

There’s even one where Bob Ross destroys his studio, turns into some other dude, crashes through the roof, and then a tank rolls in??

And then, finally, there’s this one… (no spoilers, you gotta watch it yourself).

Here’s what you need to know about AI today:

• A federal judge allowed a copyright lawsuit against AI art generators.
• Google quietly released Imagen 3, its “ethical” new AI image model.
• Runway released Gen-3 Alpha Turbo, which makes ultra-fast AI videos.
• A study found that students using chatbots for math prep performed worse on tests.

Advertise in The Neuron here.

A judge ruled that artists can go ahead with their lawsuit against Midjourney, Stable Diffusion, and Runway ML.

While Grok-2’s near-limitless image generator began wreaking havoc on X, another major piece of AI image creation news broke this week:

A federal judge allowed a landmark copyright lawsuit to proceed against AI art generators (Stability AI, Runway, Midjourney, and DeviantArt).

• The case centers around the LAION dataset, a collection of 5B images that Stable Diffusion scraped to train its image generator.
• Artists claimed Stable Diffusion invokes copyright infringement, which the defense disputes.
• Dismissed once before, the judge now agrees with the artists, allowing them to take the AI companies to court.

These are the legal questions: Does using copyrighted materials to train AI systems constitute infringement, or is it “fair use”, in the same way humans can mimic or remix styles they’ve seen?

The case has the most implication for Stable Diffusion and Midjourney—Midjourney can replicate the styles of 4,700 artists by name, and Stable Diffusion's CEO admitted it could recreate training images.

Midjourney 6.1’s rendition of an AI generator on trial.

There might be another way, too. Multiple startups are creating marketplaces for artists to license their copyrighted works, whether that’s:

• A celebrity’s likeness.
• An actor’s voice.
• A writer’s published works.
• A company’s patents.
• Or pretty much anything companies want to license.
• And yes, there’s one for art, too.  

There are so many companies doing this, they even formed their own trade group.

And that’s not even counting the major licensing deals OpenAI and Google are signing with publishers, record labels, and soon, film production companies to train new AI models.

As one licensing platform CEO said, “we’re in the Napster era of generative AI.”

FROM OUR PARTNERS

Keep your social security number off the dark web.

Every day, data brokers profit from your sensitive info—phone number, DOB, SSN—selling it to the highest bidder.

And who’s buying?

• Best case: companies targeting you with ads.
• Worst case: scammers and identity thieves.

If you want to protect your sensitive data, you need Incogni. It scrubs your personal data from the web, confronting the world’s data brokers on your behalf.

And unlike other services, Incogni removes your sensitive information from all broker types.

Protect yourself from identity theft, spam calls, and health insurers raising your rates.

Readers of The Neuron get 55% off Incogni using code NEURON. Sign up here.

Around the Horn.

Combining two new releases in one example - love it!

• Google released its new AI image model, Imagen 3 (try it here), and some have already criticized it for being “too censored.”
• Runway’s Gen-3 Alpha Turbo is out, and you can now make AI videos faster than you can type.
• Many publishers can’t afford to block Google’s AI scrapers from summarizing their content in the SGE, as they risk not showing up in its search results at all.  
• A new study found that students who use chatbots to prepare for math tests perform worse on the tests.

Treats To Try.

Tavus offers tools to create AI digital twins that can speak, see, and hear in real-time.

1. See our top 51 AI Tools for Business here!
2. Photo AI ​​generates realistic AI photos and videos of products, replacing traditional photoshoots (check out this example).
3. Gigabrain is a search engine custom-built to search Reddit for useful posts (kinda like Perplexity, but just for Reddit).  
4. Canny centralizes and organizes customer feedback to help companies make data-driven product decisions.
5. ​​postgres.new lets you create and experiment with databases instantly in your web browser (using AI help to make it easier for non-technical users).
6. Every released Sparkle, a tool to organize your Downloads, Desktop, and Documents folders (only for Mac users for now).

*We have select secondary ad spots available in August. Get your brand in front of 450,000+ professionals here.

Intelligent Insights

• Check out this deep dive on what Grok-2 is actually good at (besides wild images).
• Watch a journalist test out Gemini Live, live.
• Turns out, today’s text generators can output a lot longer texts than we thought.
• Great recap from Ethan Mollick on how far we’ve come in AI development over the last 21 months.
• Claude released a big new feature—here’s a brief explanation of why it’s useful.
• Read what filmmakers think about the limitations (and possibilities) of video generators.
• Listen to this interview on the difference between memory and intelligence when it comes to AI (be warned, it’s very nerdy!)—there’s a short thread on the topic, too.

A Cat's Commentary.