Microsoft

A one-prompt attack that breaks LLM safety alignment

As LLMs and diffusion models power more applications, their safety alignment becomes critical. Our research shows that even minimal downstream fine‑tuning can weaken safeguards, raising a key question ...
IEEE

A Review of Safe Reinforcement Learning Methods for Modern Power Systems

Abstract: Given the availability of more comprehensive measurement data in modern power systems, reinforcement learning (RL) has gained significant interest in ...
CIO

From systems of record to systems of work

If you’re using AI only to automate, you’re already behind. The real winners are rebuilding work itself with systems designed to learn. Every CEO I talk to has an AI task force. Most have pilots ...
Hosted on MSN

Dangerous behavior safety examples

UPS is firing its biggest customer -- and Wall Street finally understands why I was a combat soldier in Iraq. Here's the 1 question everyone should be asking about ICE right now Latest on snow chances ...
IEEE

Safe Control Synthesis for Neural Network Control Systems via Constrained Zonotopes

Abstract: This letter addresses the safe control synthesis problem for neural network control systems subject to bounded unknown disturbances and known exogenous inputs. A forward reachability ...