OpenAI is training models to 'confess' when they lie - what it means for future AI

A new study made a version of GPT-5 Thinking admit its own misbehavior. But it's not a quick fix for bigger safety issues.
Interesting Engineering

DoorMan: Humanoid robot trained in new system beats human operators at opening doors

A simulation-trained DoorMan system helps a Unitree G1 outperform human operators in door opening speed and reliability.
IEEE

Trajectory Tracking with Zero-Shot Reinforcement Learning on Aerial Robots: A Real-World Evaluation

Abstract: Reinforcement learning (RL) has demonstrated the ability to perform precise control in aerial robot applications. However, RL policies have struggled with the sim-real gap, which often ...