AI IP Battles

This is The Startup Breakdown, the newsletter where we learn, laugh, and love startups. By joining this growing community of hundreds of future startup aficionados (think i spelled that right?), you're getting a beachside view of the ocean that is the startup and VC scene. This ain’t your grandpa’s newsletter, so prepare yourself for an inbox full of 4/20 jokes and Succession references.

Happy Wednesday, folks.

We’ve made some insane progress at Infobot lately, and I’m excited to tell you that we’re about ready to tell you more about it.

Info is officially available in the App Store!

We’re in an all out sprint to improve our content, so we’re looking for power users (preferably those in the Bay Area for now) who would be willing to join us and help us to build the coolest news product in the world.

If you want to play a critical role in building this product, download the app from the App Store now!

AI IP Battles: OpenAI's Court Victory & Reddit's $60M Content Deal Unfold

It was an interesting week of IP news in the AI space.

On the positive side (at least for OpenAI), a judge rejected the case from various authors such as Sarah Silverman and Paul Tremblay who accused OpenAI of “grifting” their work to train ChatGPT.

It was determined that the authors’ claims weren’t valid as ChatGPT’s output was not similar enough to the works in question. They also failed to convince the judge that OpenAI had removed credit material from the training data.

The judge didn’t reject the “core claim” that OpenAI had violated copyright laws in using their work for training, though the non-profit expects to defeat the suit in full in the near future.

Not great news for the many creatives out there concerned about their blood, sweat, and tears being swiped to make ChatGPT more artistic.

Not all news was good news for OpenAI, though. The U.S. Patent and Trademark Office denied OpenAI’s attempts to trademark "GPT" and "ChatGPT."

This sucks for Sam Altman and the folks trying to get rid of all of the “GPT” copycats and grifters, it’s not the end of the world. ChatGPT and GPT are practically synonymous with OpenAI’s flagship product at this point.

Finally, Reddit is gearing up for a public listing, and as investors search for a path to monetization, the news that the platform reached a recurring $60 million content licensing deal with an undisclosed major AI company is massive.

While it’s not clear which company it is (lean towards Google given its demonstrated willingness to license content), this provides Reddit a clear path towards monetizing one of the most unique products on the internet and sets a precedent for how social data may be used to improve the next generation of models that aren’t already associated with a particular AI company (Grok and Twitter).

Coming on the heels of the NYT lawsuit, the Wild West scraping days are over, and content expenses will have to be factored into AI business models from day one.

Few “startups” will have to worry about this. If the costs to entry were astronomical before, a new space-related term will have to be created to describe how expensive building an “AI” company is now.

However, it does open a new monetization strategy for free apps with valuable content. Data is invaluable, and though the privacy and data collection crowds will object, ads might not be the lone path to building a profitable business.

The battle for who owns what in the world of LLMs is far from over. This next year will be critical in the evolution of the industry as a whole.

OpenAI successfully defended itself in a lawsuit over IP claims, with a judge finding the use of authors' works to train ChatGPT not in violation of copyright laws. Meanwhile, Reddit announced a significant $60 million content licensing deal with an undisclosed AI company, showcasing the growing commercial and legal complexities in the AI industry.

Have other feedback? Reply directly to this email and let me know!

Cheers to another day,

Trey