IEEE

Reward Hacking in Reinforcement Learning and RLHF: A Multidisciplinary Examination of Vulnerabilities, Mitigation Strategies, and Alignment Challenges

Abstract: Reinforcement Learning (RL) agents optimize policies based on provided rewards, yet may exploit unintended loopholes in the reward design, a phenomenon known as reward hacking. With the rise ...
GitHub

Sparse Autoencoders Learn Monosemantic Features in Vision-Language Models

Sparse Autoencoders (SAEs) have recently gained attention as a means to improve the interpretability and steerability of Large Language Models (LLMs), both of which are essential for AI safety. In ...
Forbes

The Language Hack That Could Fix Your Work Stress

This voice experience is generated by AI. Learn more. This voice experience is generated by AI. Learn more. Workplace metaphors like "crush deadlines" and "go to war" trigger stress responses before ...
ssd.eff.org

Government Hacking and Subversion of Digital Security

Protect digital privacy and free expression. EFF's public interest legal work, activism, and software development preserve fundamental rights. DONATE TO EFF ...
IEEE

Using Large Language Models to Extract UML Class Diagrams from Java Programs

Abstract: Many organizations rely on software systems to perform their core business operations. These systems often require modernization to accommodate new requirements and demands over time. Visual ...