NIST data protection guidance, international standards follow Biden executive order

Following Joe Biden’s executive order for data protection at the end of February, the Department of Commerce has announced three draft guidance documents from the National Institute for Standards and Technology, a draft plan for international standards, and a new measurement program which is open to comments from the public. The General Services Administration (GSA) has published its own guidance document, specific to procuring generative AI and other specialized technologies.

NIST’s AI RMF Generative AI Profile (NIST AI 600-1) can help organizations identify the risks posed by generative AI. It explores 13 risks and over 400 actions developers can take to mitigate them. Some risks include lower barriers to entry for hacking, phishing, malware, and other cybersecurity attacks, and the production of hate speech.

The Secure Software Development Practices for Generative AI and Dual-Use Foundation Models (NIST Special Publication (SP) 800-218A) will address malicious data training that affects generative AI systems.

NIST’s draft publication, Reducing Risks Posed by Synthetic Content (NIST AI 100-4) outlines methods for detecting, authenticating and labeling synthetic content, such as digital watermarking and metadata recording. The report provides technical ways to reduce the risks of synthetic content and improve the content’s transparency.

A Plan for Global Engagement on AI Standards (NIST AI 100-5) sets out the agency’s plan to lead the development of a global standard for AI and facilitate collaboration and information sharing.

The GSA has also released a Generative AI and Specialized Computing Infrastructure Acquisition Resource Guide. It goes over procurement, acquisition methods, and specialized computing infrastructure for generative AI.

In addition to the guides, the NIST GenAI Challenge is a program that will provide a platform for testing and evaluations for research in generative AI. It will facilitate the development of deepfake detection and promote the development of tech for identifying fake or misleading information.

NIST GenAI seeks to help people determine if a given text, image, video, or audio recording is of a genuine human or AI provenance. Both Generator and Discriminator tracks are included in the project, which is scheduled to launch as a pilot this month. A registration period is due to open in May, with an evaluation round 1 submission deadline planned for August.

The U.S. Patent and Trademark Office (USPTO) is also publishing a request for public comment on how AI could affect whether an invention is patentable under U.S law and what constitutes art.

Article Topics

GAO (Government Accountability Office)  |  generative AI  |  NIST  |  standards  |  synthetic data  |  U.S. Government