AI Inside OpenAI Sora 2 Ignites Likeness Debate
Oct 1, 2025
Jason and Jeff dive into OpenAI's latest Sora 2 update, exploring its new audio video sync and copyright opt-out options. They discuss the future of AI in video with DeepMind’s vision, and the implications of using AI-generated likenesses, especially with the controversy surrounding AI actress Tilly Norwood. Spotify's efforts to eliminate spammy tracks and California's comprehensive AI regulations also make headlines. The duo even speculates on whether mass-produced AI podcasts can truly engage audiences.
AI Snips
Chapters
Books
Transcript
Episode notes
Control Your Likeness Proactively
- Treat Sora 2 as a platform for generated content only, since shared items can't come from your camera roll.
- Monitor how your likeness appears and demand removal if creations cross personal or professional boundaries.
Video Models Becoming General-Purpose Vision Models
- Video foundation models (like DeepMind's Veo/VO3) are evolving toward zero-shot visual reasoning similar to LLMs.
- Concepts like "chain of frames" and world models suggest video AI will soon generalize across visual tasks.
Tilly Norwood Sparks Union Backlash
- The creators introduced Tilly Norwood, an AI actress, and SAG-AFTRA condemned it for using performers' work without permission.
- Jeff Jarvis compared AI characters to animated or CGI creations but questioned whether they can match human reputations built over lifetimes.
