Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack

Agent benchmarks have become the de facto measure of frontier AI competence, guiding model selection, investment, and deployment. However, reward hacking, where agents maximize a score without performing the intended task, emerges spontaneously in frontier models without overfitting. We argue tha...

China Develops Facial Recognition Payment System with Near-Perfect Accuracy

In an intent to move one step forward from others, China is planning to launch a facial recognition payment application with near-perfect accuracy that enables users to authorize their online transactions just by showing a picture of themselves. Chinese researchers from the Chongqing-based resear...

[G-SEC 46-2009] Computer Associates multiple products arbritary code execution

Computer Associates CA Anti-Virus Multiple products - arbitrary code execution Release mode : Coordinated Reference : GSEC-46-2009 - Computer Associates multiple products RCE WWW : http://blog.g-sec.lu/2009/10/computer-associates-multiple-products.html Vendor : http://www.ca.com Status : Patched...