Pitfall or Opportunity: GenAI Legal Case Studies Revealing Practical Advice
summary
A cloud inbox assistant that can pay bills on behalf of users raises concrete risks: meta-prompt and cross-prompt injection attacks, plugins with full inbox access, and legal limits on processing customer data—watch if plans include plugin apps, detailed SLAs and indemnities, or forensic logging. A retail surveillance model reports 73% accuracy, saves millions but shows higher false positives for people of color—watch if human-review thresholds, alternative datasets, or bias audits are absent. Also watch if supply chains span high-risk geographies vulnerable to data poisoning.
AI-written summary · about
chapters (7)
- 0:00GenAI legal case studies
- 1:09Panelist introductions
- 2:35AI digital assistant threat modeling
- 19:01Multimodal AI in retail theft detection
- 36:06AI-driven cyber defense risks
- 42:49Algorithm bias and legal standards
- 49:33Q&A and closing
related talks
26mSafeguarding AI: How to Manage AI and its Potential Risks
42mForward Focus – The Impact of Artificial Intelligence
30mNavigating the New AI Security Landscape
24mTechnical Tracks
37mFundamentals of AI safety and security
25mDoes Artificial Intelligence Need a General Counsel?
18mAI Is Coming For Our Secrets
06m7 practical tips for safer AI and language models