Building a Responsible AI Governance Program
summary
Panel recommends layering AI rules onto existing privacy and procurement processes, using a risk‑and‑harms‑based approach rather than a rewrite from scratch; run one‑ or two‑day co‑design workshops with cross‑functional teams from security, legal, UX and data science and expect faster buy‑in. Use bias tools cautiously — legacy methods like SHAP and LIME misfit large LMs, while vendors offer newer options such as NVIDIA AutoBias and the Anthropic 70‑use‑case dataset; UC Berkeley maps about 150 AI‑specific risks and teams working with data from 40 countries report real gains.
AI-written summary · about
chapters (6)
- 0:00Introduction and Panelist Backgrounds
- 3:35The Role of Privacy Practitioners in AI Governance
- 17:15Building a Responsible AI Governance Program
- 24:32Stakeholders in AI Governance
- 30:57Frameworks and Standards for AI Governance
- 38:15Closing Remarks and Future Directions
related talks
33mResponsible AI in Practice: Real-World Examples and Challenges
31mPanel: Navigating the AI Policy Landscape
24m2025: year of Ai dilemma - ethics, regulations and innovations
39mResponsible AI
25mPanel: Ethics and Regulations in AI Governance
15mResponsible AI 2.0 : NO AI without RAI
26mResponsible AI-What, Why, How and Future!
30mResponsible and Ethical Governance of AI Models