Meta Mulls Over Paid Partnerships with News Publishers for AI Training Data

In a move that could mark a significant pivot for the tech giant, Meta is reportedly considering paying for access to high-quality training data to enhance its generative AI tools. This decision signals a potential deepening of ties with the news industry, a sector Meta has distanced itself from in recent years. According to insiders, internal discussions are underway about striking new paid deals with news publishers to gain more comprehensive access to news articles, photos, and video content. Such access could make Meta’s generative AI tools, like Meta AI, more effective and competitive in a rapidly evolving market.

This strategy is not entirely new but represents a departure from recent history. Meta, formerly known as Facebook, previously had agreements with news outlets, paying them to host links to their content on its platform. However, the company has not yet approached any news outlet formally about licensing content specifically for AI training. Should Meta proceed, these deals would be distinct from past ones and focused solely on acquiring data to train its AI models.

Meta’s relationship with the news industry has been rocky. Over the last 18 months, the company has significantly reduced its investment in news-related ventures, eliminating a $2 billion budget for its News division just last year. CEO Mark Zuckerberg has boasted that Meta’s own data, used for training its Llama large language model, surpasses the Common Crawl dataset, a colossal set of web-scraped data employed widely for AI training. Yet, relying solely on its own data could leave Meta’s AI outputs lagging behind rivals like Google and OpenAI, who have already secured deals with news publishers for richer content access.

The landscape of generative AI saw a seismic shift nearly two years ago with the advent of ChatGPT, which brought AI-driven chatbots into the public eye. In response, news outlets and other websites began blocking automated bots from Common Crawl and OpenAI that were scraping their content for free. The US Copyright Office is now considering new regulations to govern the use of generative AI, adding another layer of complexity to Meta’s decision-making process. Without free and constant access to updated content, Meta AI’s responses to user queries about current events could become increasingly limited, outdated, or inaccurate.

Given the competitive nature of the generative AI market, Meta’s potential partnerships with news publishers could be a game-changer. Many news publishers appear open to licensing deals, driven by the pragmatic belief that any additional revenue is better than none. A pivot back towards the news industry could provide Meta with the timely and relevant data it needs to bolster its AI capabilities, ensuring it remains a formidable player in the generative AI arena. Whether Meta will indeed take this leap remains to be seen, but the ramifications for both the tech and news industries could be profound.