The old adage that you should never get into a fight with someone who buys ink by the barrel load doesn’t resonate in the digital age as well as it used to. But there’s no doubt news publishers’ complaints about AI startups pirating their copyrighted material has drawn plenty of coverage.

There may be a solution. A growing number of startups are emerging to try and help publishers, and other content owners, figure out how to maximize the money they can make from licensing their material to AI firms. Firms like New York City-based TollBit and London-based Human Native AI have devised technological solutions to help publishers, which otherwise are stuck negotiating one-off content licensing deals that many in the industry worry will haunt publishers in future years.

What gives these startups reason to think they can help publishers make more AI-related money is that AI firms are already spending a lot of money to get data for training their large language models or making sure their LLMs can reference up-to-date information, not always legitimately. For instance, a multi-billion dollar industry of web scraping firms are now selling data they’ve scraped from publishers’ sites to some LLM developers and AI search engines.

Human Native and TollBit want to convince AI companies to stop paying scraping firms, and to spend that money paying content owners on a recurring basis, as opposed to the licensing deals now getting done, which often aren’t that generous to content owners.

The two are taking different approaches. Human Native AI is building a data marketplace where publishers can upload their content in a form that has been cleaned up and suitable for AI training. AI companies can pay a fee to access the data. Publishers get a recurring subscription fee or a cut of the AI companies’ revenue.

Meanwhile, TollBit is more focused on helping publishers and content creators understand when their content is accessed by AI companies’ bots. The startup helps news publishers create a copy of their site that bots are automatically directed to when they need to reference up-to-date information that’s not already contained in an AI model (otherwise known as “retrieval augmented generation.”) The AI companies then pay rights holders a fee every time their bots access their content.

These approaches aren’t perfect. With AI startups like Adept developing autonomous agents that can navigate the web like humans do—scrolling, making Google searches and clicking on buttons—it could get more difficult for firms like TollBit to distinguish human site visitors from bots.

And while these startups might help facilitate connections between publishers and AI companies or help publishers get paid when AI search engines reference their work to answer questions from users, none of them have a good answer for how much publishers should be charging for the content that’s used during the model training process.

That’s complicated further by contradicting signals from AI firms. Even while OpenAI shells out millions to strike deals with publishers, its CEO Sam Altman has made comments in the past like, “We actually don’t need to train on [the New York Times’] data…Any one particular training source, it doesn’t move the needle for us that much.”

Here’s what else is going on…

It’s not every day that you get an inside look into the mind of one of the earliest backers of OpenAI. Luckily for you, the latest episode of “More or Less” has exactly that, with an interview with prolific investor Vinod Khosla.

Some of the most interesting highlights include:

  • Khosla’s advice to startups to “ignore the cost of compute.” The investor believes that compute will soon become “dirt cheap” as engineers from OpenAI, Google and cloud companies work to reduce the costs of expensive AI chips. Founders that spend money optimizing their software will be wasting capital, he added.

  • No one model to rule them all. Even though Khosla was one of the earliest backers of OpenAI, his firm, Khosla Ventures, has also invested in rival Symbolica, a startup developing less compute and data-intensive AI models. So the VC doesn’t see a winner-take-all future for AI model development.

  • The AI funding frenzy is calming down. Khosla said that sky-high valuations in the AI industry have started to stabilize in the last three months. Case in point: 80% of his firm’s seed-stage investments are below a $25 million post-investment valuation, he estimated.

  • AI will be “hugely” deflationary. In a world where AI handles most of our mundane, everyday tasks and generates entertainment like music and movies, prices and salaries could look a lot different (read: lower). What does that mean for us? Not sure, but it’s an interesting version of the world to think about!

See The Information’s Generative AI Database for an exclusive list of private companies and their investors.

AccountsIQ, which offers AI-powered accounting software, raised $65 million in Series C funding from Axiom Equity.

Enveda, which aims to develop drugs from medicinal plants, raised $55 million in Series B funding from Microsoft, The Nature Conservancy, Premji Invest, Lingotto Investment Fund, Kinnevik, True Ventures, FPV, Level Ventures and Jazz Venture Partners.

Constructor, which develops AI-powered search software, raised $25 million in Series B funding led by Sapphire Ventures with participation from Silversmith Capital Partners.

Autify, an AI agent developer, raised a $13 million Series B funding round co-led by Global Capital Partners and LG Technology Ventures, with participation from World Innovation Lab, Salesforce Ventures, Archetype Venture and Uncorrelated Ventures.

GPTZero, which detects AI-generated content, raised $10 million in Series A funding led by Footwork co-founder Nikhil Basu Trivedi.

Zeta Labs, an autonomous AI agent startup, raised $2.9 million in pre-seed funding led by Daniel Gross and Nat Friedman, with participation from Earlybird VC, Kaya VC and others.

New From Our Reporters

By Jing Yang China’s Didi Makes Its Pitch to U.S. Investors Ahead of Potential 2025 HK IPO

By Julia Black The Top 5 The Very Best Wedding Planners for a Tech Mogul Blowout

By Aaron Holmes, Natasha Mascarenhas and Julia Hornstein Exclusive OpenAI CEO Says Company Could Become Benefit Corporation Akin to Rivals Anthropic, xAI

Recommended Newsletter

100k+ readers rely on The Information’s Creator Economy newsletter for coverage of the creator startups making waves, big tech companies’ social media playbooks, and scoops on the sector’s biggest hires. Start receiving the free newsletter here.

Upcoming Events

More details Tuesday, July 16—There at the Beginning: The U.S.-China Battery Showdown
How did the U.S.-Chinese electric vehicle and battery war begin? And can the U.S. still possibly win? For an unusually incisive and entertaining live chat, The Information welcomes Jeff Chamberlain, CEO of Volta Energy Technologies, a venture capital firm investing in next-gen batteries, and a key figure in the start of the U.S.-China geopolitical rivalry.

More details Thursday, September 5—The Information’s San Francisco AI Summit
The Information’s AI Summit, taking place in September 2024 in San Francisco will bring attendees to the forefront of this fast-moving industry and all the questions that come along with it. Space is limited. Tickets go on sale soon.

More details Monday, October 7—The Information’s 2024 WTF Summit
Save the date: The premier event for women in tech media and finance is heading back in Napa Valley from 10/7-10/8. Connect with top professionals, and gain insights on AI, fintech, media and investing. To get your brand involved, contact [email protected].

Thank you for reading the AI Agenda Newsletter! I’d love your feedback, ideas and tips: [email protected].

If you think someone else might enjoy this newsletter, please pass it forward or they can sign up here.