{"slug": "researchers-can-now-reverse-engineer-llm-prompts-from-output-text-with-near", "title": "Researchers can now reverse-engineer LLM prompts from output text with near-perfect accuracy", "summary": "Researchers at IIT Bombay and Adobe Research have developed an inverse language model called 'Previous-Token Prediction' that reconstructs the original prompt from an LLM's output with near-perfect accuracy, without needing access to model weights and working across different models. This poses a serious security risk for companies relying on proprietary system prompts.", "body_md": "Researchers at IIT Bombay and Adobe Research have built an inverse language model that reconstructs the original prompt from an LLM's output with near-perfect accuracy. Their method, called \"Previous-Token Prediction,\" doesn't need access to model weights and works across different models. For companies relying on proprietary system prompts, this could be a serious security risk.\n\nThe article [Researchers can now reverse-engineer LLM prompts from output text with near-perfect accuracy](https://the-decoder.com/researchers-can-now-reverse-engineer-llm-prompts-from-output-text-with-near-perfect-accuracy/) appeared first on [The Decoder](https://the-decoder.com).", "url": "https://wpnews.pro/news/researchers-can-now-reverse-engineer-llm-prompts-from-output-text-with-near", "canonical_source": "https://the-decoder.com/researchers-can-now-reverse-engineer-llm-prompts-from-output-text-with-near-perfect-accuracy/", "published_at": "2026-08-12 17:32:59+00:00", "updated_at": "2026-08-12 17:49:05.878102+00:00", "lang": "en", "topics": ["large-language-models", "ai-research", "ai-safety"], "entities": ["IIT Bombay", "Adobe Research", "Previous-Token Prediction"], "alternates": {"html": "https://wpnews.pro/news/researchers-can-now-reverse-engineer-llm-prompts-from-output-text-with-near", "markdown": "https://wpnews.pro/news/researchers-can-now-reverse-engineer-llm-prompts-from-output-text-with-near.md", "text": "https://wpnews.pro/news/researchers-can-now-reverse-engineer-llm-prompts-from-output-text-with-near.txt", "jsonld": "https://wpnews.pro/news/researchers-can-now-reverse-engineer-llm-prompts-from-output-text-with-near.jsonld"}}