Back
EPIC: Comments to NIST on Managing the Risks of Misuse with AI Foundation Models
webThis is a formal regulatory comment from EPIC to NIST, relevant to those tracking civil society input into U.S. AI governance frameworks, particularly around foundation model risk management standards.
Metadata
Importance: 42/100policy briefprimary source
Summary
EPIC (Electronic Privacy Information Center) submitted formal comments to NIST addressing the risks of misuse associated with AI foundation models, advocating for stronger regulatory frameworks and risk management practices. The comments focus on how large-scale AI models can be exploited for harmful purposes and what governance mechanisms should be established. EPIC argues for accountability measures, transparency requirements, and meaningful oversight of foundation model development and deployment.
Key Points
- •EPIC calls for NIST to establish robust misuse risk management guidelines specifically tailored to foundation models and their broad downstream applications.
- •Comments emphasize the need for transparency and accountability from developers of large-scale AI models regarding potential harms and misuse vectors.
- •EPIC advocates for privacy protections and civil liberties considerations to be central to any NIST framework on foundation model risk.
- •The submission highlights that foundation models' general-purpose nature creates unique dual-use risks that require proactive governance rather than reactive measures.
- •EPIC urges NIST to coordinate with other agencies and stakeholders to ensure consistent, enforceable standards rather than purely voluntary guidelines.
Cited by 1 page
| Page | Type | Quality |
|---|---|---|
| NIST and AI Safety | Organization | 63.0 |
Cached Content Preview
HTTP 200Fetched Apr 9, 202622 KB
EPIC Comments to NIST on Managing the Risks of Misuse with AI Foundation Models – EPIC – Electronic Privacy Information Center
Join EPIC’s fight to STOP THE SURVEILLANCE STATE.
Defend Privacy. Protect Democracy. Support EPIC.
epic.org/stop-the-surveillance-state
Dismiss message.
APA Comments
EPIC Comments to NIST on Managing the Risks of Misuse with AI Foundation Models
2024-09824
DOWNLOAD EPIC Comments on NIST AI 800-1 09.09.24 pdf 243.4KB
Contents
Contents
Introduction
I. Sociotechnical AI Risks Are Misuse Risks
II. Preserving Consumer Privacy is Crucial to Managing the Misuse Risks of Foundation Models
Conclusion
COMMENTS OF THE ELECTRONIC PRIVACY INFORMATION CENTER
to the
NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY
Request for Comments on the U.S. Artificial Intelligence Safety Institute’s Draft Document: Managing Misuse Risk for Dual-Use Foundation Models
No. 2024-17614
September 9, 2024
Introduction
The Electronic Privacy Information Center (EPIC) submits these comments in response to the National Institute of Standards and Technology’s (NIST’s) Request for Comments on the U.S. Artificial Intelligence Safety Institute’s Draft Document on Managing Misuse Risk for Dual-Use Foundation Models (NIST AI 800-1). 1
EPIC is a public interest research center in Washington, D.C., established in 1994 to secure the fundamental right to privacy in the digital age for all people through advocacy, research, and litigation. 2 We advocate for a human-rights-based approach to AI policy that ensures new technologies are subject to democratic governance. 3 Over the last decade, EPIC has consistently advocated for the adoption of clear, commonsense, and actionable AI regulations across the country. 4 EPIC has also published extensive research on emerging AI technologies like generative AI, 5 as well as the ways that government agencies develop, procure, and use AI systems around the country. 6 EPIC is a member of NIST’s U.S. Artificial Intelligence Safety Institute Consortium (AISIC).
As the U.S. Artificial Intelligence Safety Institute (AISI) considers updates to Draft Document NIST AI 800-1, Managing Misuse Risk for Dual-Use Foundation Models, EPIC reemphasizes our call for NIST and its affiliated entities to implement actionable AI risk mitigation strategies with strong incentive structures and accountability mechanisms—steps that will ensure that AI developers and deployers faithfully adopt the practices and implementation recommendations within NIST AI 800-1. 7 At the same time, EPIC encourages AISI to view the misuse risks of generative AI technologies, including dual-use and multimodal foundation models, as extensions of traditional AI and automated decision-making risks, rather than qualitativ
... (truncated, 22 KB total)Resource ID:
e2d1acdd85129b22 | Stable ID: sid_dwlYYwX9Jk