AI Vulnerabilities and Global Governance: Navigating the Future

Explore the latest revelations in AI security as adaptive attacks expose vulnerabilities in large language models, and uncover the bias in the UK's AI system for detecting benefits fraud. Delve into Midjourney's decision to block images of Biden and Trump, highlighting the ethical challenges of AI in politics. Plus, learn about the U.S.-led UN resolution aiming to ensure AI is safe and accessible globally, setting a new standard for international cooperation in technology governance.

Sources:
https://www.marktechpost.com/2024/12/08/adaptive-attacks-on-llms-lessons-from-the-frontlines-of-ai-robustness-testing/
https://www.theguardian.com/society/2024/dec/06/revealed-bias-found-in-ai-system-used-to-detect-uk-benefits
https://apnews.com/article/midjourney-ai-imagegenerator-biden-trump-deepfakes-bc6c254ddb20e36c5e750b4570889ce1
https://apnews.com/article/un-artificial-intelligence-resolution-rules-governance-goals-b442f2701139780526b34ba0527e9425

Outline:
(00:00:00) Introduction
(00:00:40) Adaptive Attacks on LLMs: Lessons from the Frontlines of AI Robustness Testing
(00:03:29) Revealed: bias found in AI system used to detect UK benefits fraud
(00:06:06) AI image-generator Midjourney blocks images of Biden, Trump
(00:08:55) US spearheads first UN resolution on artificial intelligence