Scale AI exposes confidential AI training documents
Scale AI—a company trusted by Google, Meta, and xAI—accidentally left thousands of sensitive files open to the public through unsecured Google Docs links.
The leaked info included project guidelines, chatbot details, and even personal contractor data.
This comes just after Meta invested $14.3 billion for a big stake in the company.
Confidential details about Google Bard and xAI's Project Xylophone leaked
At least 85 documents were out in the open, showing off confidential details about Google's Bard chatbot, xAI's "Project Xylophone," and Meta's training files.
Even emails and pay disputes from contractors were visible—and some spreadsheets could be edited by anyone with the link.
Major partners rethinking their relationships with Scale AI
Scale AI has now switched off public sharing and started an investigation.
They're telling clients they take security seriously, but major partners like Google and OpenAI are rethinking their relationships—especially since Meta now owns nearly half the company.
Incident shows how high the stakes are for everyone involved
This leak highlights how even big tech partners can slip up on data security.
With so much competition in AI right now, companies are under pressure to keep things safe—and this incident shows just how high the stakes are for everyone involved.