Toward AI Security.pdf

发布时间：2022-06-06 发布人：admin 分类：说明书资料大小：9.33M 资料格式：pdf 举报版权申诉

zongzhiyuan-12122373-4744300845217670289.pdf-第1页.png

第1页 / 共94页

zongzhiyuan-12122373-4744300845217670289.pdf-第2页.png

第2页 / 共94页

zongzhiyuan-12122373-4744300845217670289.pdf-第3页.png

第3页 / 共94页

zongzhiyuan-12122373-4744300845217670289.pdf-第4页.png

第4页 / 共94页

zongzhiyuan-12122373-4744300845217670289.pdf-第5页.png

第5页 / 共94页

zongzhiyuan-12122373-4744300845217670289.pdf-第6页.png

第6页 / 共94页

zongzhiyuan-12122373-4744300845217670289.pdf-第7页.png

第7页 / 共94页

zongzhiyuan-12122373-4744300845217670289.pdf-第8页.png

第8页 / 共94页

文本预览

C E N T E R F O R L O N G - T E R M C Y B E R S E C U R I T Y C L T C W H I T E P A P E R S E R I E S Toward AI Security G L O B A L A S P I R A T I O N S F O R A M O R E R E S I L I E N T F U T U R E J E S S I C A C U S S I N S N E W M A N

C L T C W H I T E P A P E R S E R I E S Toward AI Security G L O B A L A S P I R A T I O N S F O R A M O R E R E S I L I E N T F U T U R E J E S S I C A C U S S I N S N E W M A N FEBRUARY 2019 C E N T E R F O R L O N G - T E R M C Y B E R S E C U R I T Y

iviv TOWARD AI SECURITY

Contents AI Security Map 13 1. Digital / Physical Domain 16 2. Political Domain 20 3. Economic Domain 25 4. Social Domain 29 Acknowledgements 2 Abstract 3 Recommendations 4 Introduction 6 AI 101 8 The Transformative Nature of AI 11 AI Security 13 Global AI Policy Responses 34 Global AI Security Priorities 68 Working with the Private Sector 70 Conclusion 72 Appendix I: AI Policy Compendium 73 Appendix II: Overview Of AI Policy Interests From Ten Countries 76 Endnotes 78 China 36 France 39 United Kingdom 45 United States 51 Canada 57 India 59 Japan 60 Singapore 62 South Korea 63 United Arab Emirates 65 1 TOWARD AI SECURITYTOWARD AI SECURITY

Acknowledgements The Center for Long-Term Cybersecurity (CLTC) would like to thank the following individuals for their expertise, feedback, and contributions to this document: Anthony Aguirre, Seth Baum, Niel Bowerman, Jared Brown, Ryan Calo, Betsy Cooper, Elsa Kania, Jade Leung, Nicolas Miailhe, Deirdre Mulligan, Michael Page, Philip Reiner, Stuart Russell, and Helen Toner. Special thanks to Steven Weber, Ann Cleaveland, and Chuck Kapelke of CLTC for their support, feedback, and contributions to this report. CLTC would also like to thank the Hewlett Foundation for making this work possible. 22 TOWARD AI SECURITY

Abstract This report uses the lens of global AI security to investigate the robustness and resiliency of AI systems, as well as the social, political, and economic systems with which AI interacts. The report introduces a framework for navigating the complex landscape of AI security, visualized in the AI Security Map. This is followed by an analysis of AI strategies and policies from ten countries around the world within this framework to identify areas of convergence and diver- gence. This comparative exercise highlights significant policy gaps, but also opportunities for coordination and cooperation among all surveyed nations. Five recommendations are provided for policymakers around the world who are hoping to advance global AI security and move us toward a more resilient future. The steps nations take now will shape AI trajectories well into the future, and those governments working to develop global and multistakeholder strategies will have an advantage in establishing the international AI agenda. 3 TOWARD AI SECURITYTOWARD AI SECURITY

Recommendations Based on the analysis of the gaps and opportunities in national AI strategies and policies, we provide five recommendations for policymakers hoping to harness and direct AI technologies for a more resilient and beneficial future. These recommendations outline concrete actions that can be taken now to address a complex and quickly changing sociotechnical landscape: 1. Facilitate early global coordination where common interests can be identified. As autonomous systems become more ubiquitous and capable, their reach and effects will be more consequential and widespread. Global coordination and cooperation will be essen- tial for ensuring sufficient oversight and control, but such cooperation will be harder to achieve the longer we wait due to technological and institutional “lock-in”. The numerous areas of convergence identified in this report can be leveraged as opportunities for collab- oration and innovation, sharing best practices, and preventing global catastrophic risks. 2. Use government spending to shape and establish best practices. Governments have an opportunity to establish standards and best practices while promoting AI development and use, for example by implementing guidelines for government procurement of AI systems, and by adding criteria such as safety, robustness, and ethics to AI R&D funding streams. Additionally establishing processes to support transparent and accountable government funding and use of AI technologies will help prevent misuse throughout public services and protect government actors from the limitations and vulnerabilities of AI tools. 3. Investigate what is being left on the table. The landscape of AI security is broad and complex, as indicated in the AI Security Map presented in this report. The analysis of policy documents identifies many gaps in different nations’ current AI policy approaches. Govern- ments may choose to prioritize a sub-set of issues, but they should recognize the opportu- nities and challenges they could be neglecting. 4. Hold the technology industry accountable. Many governments rightfully emphasize the importance of partnership and engagement with industry and other AI stakeholders. How- ever, while some firms are addressing AI challenges, significant gaps remain. Policymakers have the unique primary responsibility to protect the public interest, and this responsibility carries even greater weight during periods of significant technological transformation. 44 TOWARD AI SECURITY

分享到：

赞收藏

资料库

Toward AI Security.pdf

相关推荐

安全技术

热门标签

最新资料