The U.S. Department of Commerce’s National Institute of Standards and Technology (NIST) has launched an initiative to gather public feedback on the safe and responsible use of artificial intelligence (AI). This effort is in response to a presidential executive order and seeks input up until February 2, 2024. NIST's goal is to collect essential information to aid in testing and ensuring the safety of AI systems.
Commerce Secretary Gina Raimondo highlighted that this initiative stems from President Joe Biden's recent executive order. The directive tasks NIST with creating guidelines for AI assessment, fostering standards, and setting up an environment to test AI systems. This framework aims to assist the AI community in developing technology that is safe, reliable, and ethically responsible.
NIST's request specifically calls for insights from AI companies and the general public on managing risks associated with generative AI, particularly in minimizing the spread of AI-generated misinformation. Generative AI, known for creating text, images, and videos from broad prompts, has been a source of both excitement and concern, raising issues such as job displacement, election interference, and the fear of AI surpassing human abilities with potentially dire outcomes.
Additionally, the request focuses on identifying key areas for "red teaming" in AI risk assessment and establishing best practices for it. Red teaming, a concept originating from Cold War simulations, involves a group simulating adversarial actions to identify vulnerabilities in systems, processes, or organizations. This technique, widely used in cybersecurity, is crucial for uncovering new risks in AI.
In August, the U.S. hosted its first public red team event focused on AI at a cybersecurity conference, organized by AI Village, SeedAI, and Humane Intelligence. Following this, in November, NIST announced the creation of a new AI consortium. This consortium, requiring qualified applicants, aims to develop policies and measures that ensure U.S. legislation adopts a human-centered approach to AI safety and governance.
