Zifan Peng

Abliterated LLMs can dramatically enhance vulnerability detection and patch validation, achieving up to 67.8% usability in early-stage validation compared to just 29.9% for their aligned counterparts.

EvoMap, a leading agent-to-agent collaboration network, is plagued by mass-produced, unused assets, gamed scoring systems, and vacuous quality checks, revealing the dark side of unchecked, scalable growth in AI collaboration.

Current MLLMs fail to detect covert advertisements, revealing a critical gap in social media moderation that could mislead consumers and pose ethical risks.

Users often dangerously misunderstand the true scope of authority they've granted to computer-use agents, even while recognizing abstract risks.