Panel: Main legal risks and challenges with LLMs / generative AI models
summary
Panelists map concrete legal pain points: models train on scraped social media and personal data so 175 billion parameters scale leaks consent and opt-out impossible enforcement problems, GDPR rights like deletion are technically unclear, and liability likely shifts to the deployer who ships an app built on opaque foundation models. Regulators juggle EU AI Act drafts, Biden’s cybersecurity-leaning executive order, and litigation such as Clearview; open-source efforts like BigScience/Bloom (about 59 languages, ~1,000 collaborators) show transparency can reduce but not erase bias.
AI-written summary · about
chapters (12)
- 0:00Panel introduction
- 1:15Legal risks in training
- 10:06Global AI regulations
- 18:32Ria and AI Act priorities
- 22:25Challenges in regulating large language models
- 25:44Bias detection and accountability
- 30:31Haio and model biases
- 35:54AI bias consequences
- 38:49Q&A on accountability
- 44:52AI vs microprocessor regulation
- 48:10Adapting regulations for AI
- 52:27Regulating open source AI
related talks
56mEthics in Machine Learning Panel
01h 06mPanel on Interpretability in the Law
31mPanel: Navigating the AI Policy Landscape
01h 09mGOVERNANCE AND REGULATION OF AI FROM THE PERSPECTIVE OF AUTONOMY AND PRIVACY
33mEthics in Machine Learning Panel Q&A
41mAI and the Law
01h 08mTHE GOVERNANCE OF AI: CONVERGENCE OR DIVERGENCE?
01h 11mThe Intersection of AI and Regulation How Organisations and Regulators Should...