At the 2023 Defcon hacker conference, Humane Intelligence collaborated with AI tech firms, transparency groups, and US government support for a 'red-teaming' effort targeting generative AI platforms. This initiative aims at exposing vulnerabilities and enforcing rigorous scrutiny of influential AI systems, extending to a nationwide red-teaming event with NIST to evaluate AI office productivity software. Participants, including those from the general public, will test the security, resilience, and ethics of these AI technologies, with the end goal of democratizing the evaluation process and ensuring these models meet user needs.
Source: https://www.wired.com/story/nist-humane-intelligence-generative-ai-red-team-contest/
"id": "hum000082624",
"linkid": "humaneintelligence",
"type": "Breach",
"date": "8/2024",
"severity": "25",
"impact": "1",
"explanation": "Attack without any consequences"
Join the conversation