In one of the largest purchases the social media behemoth has undertaken since its 2014 agreement to acquire WhatsApp, Meta has agreed to pay $14.8 billion for a 49 percent ownership in the artificial intelligence data company Scale AI, according to a person familiar with the situation.

An individual who spoke on condition of anonymity to discuss private company matters claimed that the deal with Scale AI, which might be revealed as early as Wednesday, will give Alexandr Wang, the CEO of the start-up, a senior position at Meta overseeing a team focused on building superintelligence.

A hypothetical technology that could outperform humans at every endeavor is referred to as superintelligence. According to the source, Meta, a current Scale AI client, plans to increase its usage of the business’s data-labeling services. The Information has previously reported on the arrangement. Meta’s decision to establish a new AI lab was previously covered by Bloomberg News and the New York Times. Scale AI and Meta chose not to comment.

Through the agreement, Meta will have greater access to Scale AI’s services, which have changed over time. The company first focused on self-driving cars and used “digital sweatshops” in places like the Philippines to link big tech corporations with human workers to categorize the vast volumes of data required to train AI models.

Scale shifted to the kinds of contractors and data needed for subsequent phases of AI research, such as giving chatbots a more human voice or assisting them in become better in particular domains, like defense or healthcare, when the technology advanced to the point where it required less labeling.

Scale’s contractor hiring platform, Outlier AI, posted 28 job openings on Tuesday, primarily for specialized human trainers with expertise in advanced biology, chemistry, and physics, as well as many voice-training positions in Chinese and Arabic.

William Kovacic, a former chair of the Federal Trade Commission and a law professor at George Washington University, stated that antitrust regulators in the Trump administration may scrutinize Meta’s acquisition. They expressed skepticism about the influence and power of large technology companies, such as Meta, but were cautious about regulating AI.

Earlier this year, despite Meta CEO Mark Zuckerberg’s efforts to persuade the White House to reach a settlement, the FTC decided not to abandon its case alleging that Meta had acquired Instagram and WhatsApp.

Panelists at an FTC event earlier this month demanded strict regulatory measures to help make the internet safer for children and teenagers. “AI may pose a much-needed competitive and innovative challenge to incumbent Big Tech firms,” however, according to FTC Chair Andrew Ferguson’s September statement.

He continued by saying that a hasty legislative response will simply hinder innovation, strengthen the position of Big Tech incumbents, and force AI pioneers to relocate to countries that are more hospitable to them but possibly antagonistic to the US.

According to Kovacic, if antitrust authorities look into Meta’s acquisition, they will need to decide if the business qualifies as a dominant force in the AI space before taking any action.

According to the business community’s assessment, “I feel that Meta is not leading the field in the development of artificial general intelligence — that it is lagging behind,” he stated.

As the social media behemoth tries to compete with other companies that are more renowned in generative AI, such OpenAI, Anthropic, Google, Microsoft, and China-based DeepSeek, its partnership with Wang is the most recent in a string of personnel changes. According to Zuckerberg, Meta’s AI chatbot will surpass competitors like ChatGPT and Claude to become the industry standard AI helper.

As OpenAI and Google started to publish fewer papers and disclose less about the development of their technology, Meta gained a lot of favor with industry insiders when its well-liked AI model, named Llama, was introduced in 2023.

However, Meta has made several gaffes in recent months. It was seen attempting to alter LMArena, a platform for assessing AI models, in April. Additionally, according to a Wall Street Journal report last month, Meta announced that it would postpone the release of their enormous AI model, Behemoth.

In the meantime, Meta’s newly released AI app, which Zuckerberg promoted on podcasts, appears to have failed to impress users. It has less than 10 million downloads on the Google Play Store and is ranked No. 17 in the productivity category on the Apple App Store.

Joelle Pineau, who had been in charge of Meta’s AI research lab for approximately eight years, announced her resignation in April. In order to better integrate the once autonomous artificial research lab with the company’s commercial and product ambitions, Meta made major changes during her tenure.

According to an internal letter detailing the intentions obtained by report, Meta restructured its generative AI team the next month, dividing the department into two sections devoted to consumer products and AI research.

Chris Cox, the chief product officer for Meta, wrote, “I think this structure will be a major upgrade to overcome the biggest challenges that I’ve heard from many of you and will help accelerate our overall progress.”