SB 1047, AI regulation, and unlikely allies for open models

Momentum

The bipartisan agreement that big tech companies are detrimental to Americans has fueled momentum for open-source AI. Nathan Lambert notes that this sentiment limits big tech's ability to make deals that serve their interests, while open-source AI remains agile and innovative 1. Despite the advantages big tech holds, the cultural shift towards open-weight models is promising. Lambert is optimistic about the future of open-source AI, especially with the current political climate favoring it.

Open source AI is the most mobile and fluid way to build the technology, requiring new ways of building value.

--- Nathan Lambert

The fragility of the regulatory process is evident, with many leading figures in AI expressing disapproval of bills like SB 1047. Lambert highlights that the process for clearing such laws is often flawed, influenced by political alliances rather than constituent needs 2.

   

Opportunities

Open-source AI presents significant growth opportunities despite regulatory hurdles. Lambert emphasizes the need to keep building momentum and the ecosystem for open-weight models, which have shown resilience and adaptability 1. He also discusses the potential for China to dominate the open-source AI space due to its strategic efforts and practical considerations.

The threat of China taking over control of a stack as important as open source AI by a regulation's own goal looms bigger than any national security risks of China infiltrating national labs.

--- Nathan Lambert

Lambert suggests that regulation should focus on specific areas like non-consensual deepfakes and data privacy to mitigate risks while fostering innovation. He advocates for investments in infrastructure to support researchers and ensure safe storage of data 3.