Meta Might Enter into Commercial Agreements with News Publishers to Train AI Models

In an effort to improve the effectiveness and competitiveness of its generative AI tools, such as Meta AI, in the market for generative AI search tools and chatbots provided by Google and Microsoft, Facebook parent company Meta is reportedly considering paying news organizations to better train its AI language models.

A report by Business Insider, citing two people familiar with the talks, stated that teams inside the corporation are debating internally whether to reach agreements with news publishers for increased access to news, image, and video material.

One of the sources stated, "Meta might have to pay someone." Internal discussions on content licensing are still going on, but Meta hasn't contacted any news organizations yet. Any arrangements for data access for training models would be distinct from previous contracts in which Meta compensated publishers to place links on its websites.

How Llama AI models are trained by Meta

CEO of Meta Mark Zuckerberg asserted that his business uses a bigger amount of its own data than Common Crawl, a vast collection of online scraped data that is utilized by many companies, including Meta, to train AI models.

One of the persons with knowledge of the matter stated that internal worries exist over the quality of that possessed data. For instance, postings and comments on Facebook or Instagram are not always the kind of high-quality training data that generative AI chatbots and search tools need to create quality outputs, in contrast to books, news articles, and essays.

OpenAI and News Corp. reach an agreement

The AI chatbot developer OpenAI, which also created ChatGPT, has inked a multi-year agreement with News Corp, the media conglomerate that owns publications like The Sun, MarketWatch, and The Wall Street Journal.

In order to train its AI models, OpenAI will be able to access information from more than a dozen news publishers under the terms of the agreement, which is allegedly valued at $250 million over five years.