Together AI Discloses Prompt Leak After Dragos Study
November 21, 2026
1 min read
Together says a coding agent exposed a credential for stored prompt samples. Five credentials later tested by the intruder were canaries from a Dragos experiment.... Read the full story at The Defiant