Can We Red Team Our Way to AI Accountability?

Red-teaming can only support AI accountability if laws, regulations, and enforcement to ensure protection from harm are in place.

Can We Red Team Our Way to AI Accountability? | TechPolicy.Press Can We Red Team Our Way to AI Accountability? Ranjit Singh, Borhane Blili-Hamelin, Jacob Metcalf / Aug 18, 2023 Ranjit Singh is a qualitative researcher with the Algorithmic Impact Methods Lab at Data & Society; Borhane Blili-Hamelin is the taxonomy lead at the AI Vulnerability Database, a community partner organization for the Generative Red Team (GRT) challenge at DEF CON ; and Jacob Metcalf is program director, AI on the Ground, at Data & Society. Shutterstock Republish Share Last week’s much publicized Generative Red Team (GR

related reading

• Recommendations-for-Using-Red-Teaming-for-AI-Accountability-PolicyBrief.pdfdatasociety.net
• The case for ensuring that powerful AIs are controlled — LessWronglesswrong.com
• Recommendations for Technical AI Safety Research Directionsalignment.anthropic.com
• NVIDIA AI Red Team: An Introduction | NVIDIA Technical Blogdeveloper.nvidia.com
• AI Control: Improving Safety Despite Intentional Subversion — LessWronglesswrong.com
• Defining LLM Red Teaming | NVIDIA Technical Blogdeveloper.nvidia.com
• AI Control: Improving Safety Despite Intentional Subversion — AI Alignment Forumalignmentforum.org
• Building Technology to Drive AI Governancebounded-regret.ghost.io
• GitHub - requie/AI-Red-Teaming-Guide: A comprehensive guide to adversarial testing and security evaluation of AI systems, helping organizations identify vulnerabilities before attackers exploit them.github.com
• A safe harbor for AI evaluation and red teamingnormaltech.ai
• Constitutional AI: Harmlessness from AI Feedbackarxiv.org
• [2312.06942] AI Control: Improving Safety Despite Intentional Subversionarxiv.org