top of page
AI Infosec Technical Working Group
Driving innovation in AI Information Security through prize challenges.
CLAS is a NeurIPS 2024 competition focused on improving LLM and agent safety, tackling harmful outputs and backdoor triggers to enhance AI security and reliability.
MLMAC 2023
MLMAC is a DEFCON competition focused on attributing models to their origins using only textual outputs. This challenge demonstrated the feasibility of model attribution as an emerging capability in AI Security.
Why Competitions?
01
We engage diverse solvers worldwide to address critical AI security challenges through targeted competitions.
02
Our approach eliminates barriers to entry, welcoming unexpected talent and novel solutions from all backgrounds.
03
Winners gain validation, feedback, and support to advance their innovations beyond the competition.
bottom of page