Supriti Vijay

VLoc Benchmark results establish vulnerability localization as a distinct repository-scale capability and provide a setting for studying both how security agents search for vulnerable code and when they should refrain from reporting it.

Antares-3B rivals GPT-5.5 in vulnerability localization while being 200x smaller and achieving rapid, low-cost evaluations.

FAPO achieves a remarkable +33.8 percentage point gain in performance by seamlessly transitioning from prompt optimization to structural adjustments in LLM pipelines.

LLM agents spill secrets like gossip after a few rounds, with privacy violations more than doubling in social settings compared to isolated chats.