Meta Hits Pause on $10B AI Babysitter After Digital Playground Break-In
KEY POINTS
- •Meta paused collaboration with Mercor after a recent data breach involving security vulnerabilities.
- •Mercor, valued at $10 billion in October, trains AI models using thousands of human contractors.
- •The breach related to LiteLLM forced Mercor to launch a forensic investigation while Meta declined to comment.
Meta, the company that owns everything from your newsfeed to your existential dread, pressed pause on its AI training sidekick, Mercor, after a data breach that sounds like it came straight from a hacker’s Netflix heist special. Valued at a casual $10 billion in October, Mercor’s job is babysitting AI models using thousands of human contractors who get paid presumably to teach robots not to Netflix and chill too hard. The breach traced back to LiteLLM, an open-source project whose name suggests it’s about as secure as a paper umbrella in a hurricane. Mercor, meanwhile, vowed to throw a forensic party and sorted the chaos while Meta just said ‘no comment,’ probably because 'Oops' doesn’t look good on social media.
Share the Story
(1 of 3)Source: Businessinsider | Published: 4/4/2026 | Author: Kelsey Vlamis,Charles Rollet