At the 2023 Defcon hacker conference, Humane Intelligence in partnership with AI transparency groups and US government support, initiated a red-teaming exercise targeting generative AI platforms. Inviting US residents to join the NIST's ARIA challenge, it enhances scrutiny on opaque AI systems. The exercise aims to rigorously test the security, resilience, and ethics of these technologies. This broad community participation, including traditionally underrepresented groups, will inform scientific evaluation and promote algorithmic transparency and accountability via competitive red teaming events.
Source: https://www.wired.com/story/nist-humane-intelligence-generative-ai-red-team-contest/
"id": "hum000082324",
"linkid": "humaneintelligence",
"type": "Vulnerability",
"date": "8/2024",
"severity": "25",
"impact": "1",
"explanation": "Attack without any consequences"