IIT Bombay and Adobe Researchers Develop Method to Reverse-Engineer LLM Prompts
Researchers at IIT Bombay and Adobe have created a method to reconstruct LLM prompts from output text with near-perfect accuracy, using only the generated text.
Research
190 articles
In-depth coverage of new AI research — papers, benchmarks, and breakthroughs from leading labs and academia, summarized for fast reading and grounded in the methods that matter.
LATEST NEWS
-
-
AllenAI Introduces TutorMoments Framework to Evaluate AI Tutoring Behavior
AllenAI released TutorMoments, a framework to assess if AI tutors can balance helping and letting students think, using 462 real tutoring transcripts from U.S. students.
-
Hugging Face Explores GRPO in 2026 Reinforcement Learning Training
Hugging Face highlights GRPO's role in 2026 training as frontier labs shift from preference-based to result-based rewards. OpenAI's o1 and DeepSeek-R1 exemplify this trend.
-
Researchers Find Hidden Reasoning in ChatGPT's Encrypted Thought Processes
Security researchers uncovered encrypted thought processes in ChatGPT and other models, revealing passwords and API keys in public sessions.
-
Anthropic Model Advances on Riemann Hypothesis
Anthropic's unreleased model made progress on the Riemann hypothesis, raising the lower bound of solutions for which the hypothesis holds to 650 tested ideas.
-
Google's AMIE AI Enables Real-Time Video Consultations
Google's AMIE AI system demonstrated real-time clinical video consultations in a first-of-its-kind study, with favorable assessments from evaluators and patient actors.
-
New Method Reveals AI Models' Inner Reasoning
Researchers found a way to extract hidden reasoning from AI models, showing similarities between Chinese and US models, with no conclusive proof of distillation.
-
AI Researchers Find Zoom Screen-Sharing Bug That Could Take Over Devices
Researchers from A Security discovered a Zoom vulnerability that could allow attackers to take over devices during screen-sharing calls, affecting all major operating systems.
-
IBM Unveils GridFM Neural Solver for Power Grid Optimization
IBM introduces GridFM, a neural solver that improves power grid efficiency by 15% in simulations, according to its latest research.
-
AI Unravels Complex Genetic Network Behind Schizophrenia
A study using AI identified 766 genes linked to schizophrenia, including 641 previously unknown to science.
-
Human Brain Organoids Guide Robots and Play Video Games
Scientists at UC San Diego and other institutions are using lab-grown brain organoids to guide robots through mazes and test new drugs, with some even playing video games like Pong and Doom.
-
AI Professors Navigate Challenges in Academic Research
AI researchers face growing constraints as private companies dominate large language model development, with many struggling to access resources and influence design.