Updated
Updated · O'Reilly Media · Aug 18
35 Companies Warn Washington Against Curbing Open-Source AI as Open Models Do 33% of Work
Updated
Updated · O'Reilly Media · Aug 18

35 Companies Warn Washington Against Curbing Open-Source AI as Open Models Do 33% of Work

3 articles · Updated · O'Reilly Media · Aug 18

Summary

  • An open letter backed by 35 companies including Microsoft, Palantir and Andreessen Horowitz urged Washington not to over-restrict open-source AI, arguing openness strengthens safety, cybersecurity and innovation.
  • Jensen Huang amplified the July 24 letter in his first X post, and support spread to Elon Musk and Mark Zuckerberg before the Open Secure AI Alliance launched to build shared safety tools.
  • Mozilla said open models now handle about one-third of global AI work but capture only 4% of revenue, even as 79% of developers build with them and performance gaps with proprietary systems narrow.
  • The report frames that mismatch as the real policy fight: proprietary labs want tighter control over models and surrounding software layers, while open-source backers say value and user power increasingly sit in those wrappers.
  • The dispute lands amid a broader U.S.-China split, with Washington emphasizing control for national security while Beijing pushes wider adoption of open-source AI to spread its technology globally.

Insights

If open AI handles a third of the world's work but captures almost no revenue, who will fund its future survival?
Could the hidden software layer surrounding AI models become the ultimate trap for tech dependence and global control?
As the battleground shifts from AI models to the software harnessing them, who will truly control your digital memory?