Meta's $15 Billion Scale AI Deal Could Leave Gig Workers Behind

Meta is reportedly set to invest $15 billion to acquire a 49% stake in Scale AI, in a deal that would make Scale CEO Alexandr Wang head of the tech giant’s new AI unit dedicated to pursuing “superintelligence.”

Scale AI, founded in 2016, is a leading data annotation firm that hires workers around the world to label or create the data that is used to train AI systems.

The deal is expected to greatly enrich Wang and many of his colleagues with equity in Scale AI; Wang, already a billionaire, would see his wealth grow even further. For Meta, it would breathe new life into the company’s flagging attempts to compete at the “frontier” of AI against OpenAI, Google, and Anthropic.

However, Scale’s contract workers, many of whom earn just dollars per day via a subsidiary called RemoTasks, are unlikely to benefit at all from the deal, according to sociologists who study the sector. Typically data workers are not formally employed, and are instead paid for the tasks they complete. Those tasks can include labeling the contents of images, answering questions, or rating which of two chatbots’ answers are better, in order to teach AI systems to better comply with human preferences.

(TIME has a content partnership with Scale AI.)

“I expect few if any Scale annotators will see any upside at all,” says Callum Cant, a senior lecturer at the University of Essex, U.K., who studies gig work platforms. “It would be very surprising to see some kind of feed-through. Most of these people don’t have a stake in ownership of the company.”

Many of those workers already suffer from low pay and poor working conditions. In a recent report by Oxford University’s Internet Institute, the Scale subsidiary RemoTasks failed to meet basic standards for fair pay, fair contracts, fair management, and fair worker representation.

Read More: Gig Workers Behind AI Face ‘Unfair Working Conditions,’ Oxford Report Finds

“A key part of Scale’s value lies in its data work services performed by hundreds of thousands of underpaid and poorly protected workers,” says Jonas Valente, an Oxford researcher who worked on the report. “The company remains far from safeguarding basic standards of fair work, despite limited efforts to improve its practices.”

The Meta deal is unlikely to change that. “Unfortunately, the increasing profits of many digital labor platforms and their primary companies, such as the case of Scale, do not translate into better conditions for [workers],” Valente says.

A Scale AI spokesperson declined to comment for this story. “We’re proud of the flexible earning opportunities offered through our platforms,” the company said in a statement to TechCrunch in May. 

Meta’s investment also calls into question whether Scale AI will continue supplying data to OpenAI and Google, two of its major clients. In the increasingly competitive AI landscape, observers say Meta may see value in cutting off its rivals from annotated data — an essential means of making AI systems smarter.

“By buying up access to Scale AI, could Meta deny access to that platform and that avenue for data annotation by other competitors?” says Cant. “It depends entirely on Meta’s strategy.”

If that were to happen, Cant says, it could put downward pressure on the wages and tasks available to workers, many of whom already struggle to make ends meet with data work.

A Meta spokesperson declined to comment on this story.