How do you stop AI from spreading abuse? Leaked docs show how humans are paid to write it first.
Scale AI's safety projects rely on freelancers to create graphic and "harmful" prompts. Internal documents show how they're ...