Market Rithm President Larry Ward breaks down California's new AI law regulating frontier platforms and explains why it falls short of addressing real threats like election deepfakes.
Ward highlights major loopholes in the legislation: the law only applies to platforms with over 1 million users, meaning individuals can download tools like DeepSeeker or Meta AI locally and create unregulated deepfake content. Even when watermarks exist, they're easily removable before distribution.
https://x.com/OANN/status/1981557509141020820
"There really is no heat to this law," Ward warns, noting that only AI producers—not distributors—face accountability. With the 2026 elections approaching, last-minute deepfakes could prove "really consequential" without effective enforcement mechanisms.
Key Points:
- Million-user threshold creates massive enforcement gap
- Watermarks are easily removed by bad actors
- Law targets producers, not distributors of deepfakes
- Timing concerns ahead of 2026 elections
- Local AI tools bypass regulations entirely
