AI Matters
Why it matters
Researchers reconstruct original AI prompts from output text
The Decoder Researchers can now reverse-engineer LLM prompts from output text with near-perfect accuracy
Researchers from IIT Bombay and Adobe developed a method called Previous-Token Prediction that reverse-engineers system prompts with near-perfect accuracy.
Why it matters
Companies using proprietary prompts to build applications face new security risks, as competitors can now steal their underlying instructions.
Latest brief · Aug 12, 2026