布莱切利宣言：当前人工智能安全的最佳实践.pdf

《布莱切利宣言：当前人工智能安全的最佳实践.pdf》由会员分享，可在线阅读，更多相关《布莱切利宣言：当前人工智能安全的最佳实践.pdf（17页珍藏版）》请在三个皮匠报告上搜索。

1、Thursday 13 March 2025 14:15-15:15The Bletchley Declaration:Current Best Practices for AI SafetyMaria AxenteAI Executive and AdvisorTahir LatifBoard Member and Chief AI Governance&Ethics Officer,Ethical AI AllianceBojana BellamyPresident,Centre for Information Policy Leadership(CIPL)John BowmanAI Et

2、hics Market Strategy Lead,IBMWELCOME AND INTRODUCTIONSI.Background and Contexti.AI Safety Summit(UK,2023),AI Seoul Summit(South Korea,2024)&AI Action Summit(France,2025)II.UK Perspectives on AI Safetyi.UK Government AI Opportunities Action Plan ii.ICO Pro-Growth StrategyIII.Emerging Best Practices i

3、n AI Safetyi.Threat assessment ii.AI red-teamingiii.Risk mitigationiv.Human oversightIV.Applying Data Protection Principles to Generative AIV.Best Practices in Building Accountable AI Programs VI.Questions and Answers AGENDA OUTLINEBackground&ContextParis,France10-11 February 2025Bletchley Park,UK1-

4、2 November 2023Seoul,South Korea21-22 May 2024Key themesShared visionRisk-based approach based on evidenceTransparency and explainabilityHuman oversight and controlGovernance and accountabilityInternational cooperationEthical and societal considerationsEvaluation metrics,tools for safety testing and

5、 scientific researchBletchley Declaration:OverviewThe Bletchley Declaration by Countries Attending the AI Safety Summit,1-2 November 2023AI Seoul Summit,21-22 May 2024Focus on AI safety,ethics,and responsible innovation.Progress in AI safety best practices,including threat assessment,red-teaming,ris

6、k mitigation,and human oversight.Emphasis on international collaboration and the Bletchley Declarations objectives.Seoul Declaration for Safe,Innovative and Inclusive AISeoul Ministerial Statement for advancing AI safetyAI Action Summit,10-11 February 2025Goal of accelerating AI action and adoption