Scale AI is facing backlash over massive spam and security failures while working with Google. The company, now 49 percent owned by Meta after a $14 billion deal, reportedly struggled to vet its gig workers and control fraud.
Between March 2023 and April 2024, the program called "Bulba Experts," which was supposed to staff experts to train Google’s AI, instead became flooded with "spam." This spam included gibberish, outright wrong info, and AI-generated nonsense. Many spammers got paid anyway because catching them was nearly impossible. Some even returned via VPNs after getting blocked.
The spam problem exposed a lack of background checks and loose contributor standards. A former manager said there were no degree verifications—even for projects demanding advanced qualifications—and many workers weren’t native English speakers. 
Scale AI’s use of underpaid gig contractors, mostly overseas, has drawn criticism before for creating "digital sweatshops" and wage theft allegations. Now it seems those conditions led to low-quality work, hurting projects for big clients like Google.
Attempts to fight spam included banning whole countries and using AI-detection tools like ZeroGPT. Still, thousands of spammers slipped through, dumping junk data into Google’s AI systems. This might help explain Google’s frequently criticized AI performance issues.
Scale AI denied the claims in a statement to Inc:
"This story is filled with so many inaccuracies, it’s hard to keep track."
"What these documents show, and what we explained to Inc ahead of publishing, is that we had clear safeguards in place to detect and remove spam before anything goes to customers."
The fallout highlights the serious risks of relying on poorly vetted gig labor for crucial AI training data — even when Meta is betting billions.
