Hacker News Reader

The real fundamental problem is that "digital assets" are bullshit.

An AI model is literally constructed by explicitly disrespecting copyright. The idea that a company gets to turn around and demand respect for their AI model's copyright is patently absurd.

View on HN

I feel like this can be trivially worked around by fine tuning from the initially copyrighted weights. They’re not demanding copyright, they’re just keeping them secret. If Meta keeps releasing high quality open source models, I don’t expect the closed source models to have an advantage for long.

Reply on news.ycombinator.com