All-In with Chamath, Jason, Sacks & Friedberg

The Fight Over Open Source AI, Anthropic's $1.5B Payout, NYC Socialists: Evictions = Violence?

July 24, 2026

AI Summary

5 min read

A $1.5 billion copyright settlement and a White House debate over banning Chinese open-source AI models are both symptoms of the same dynamic: the leading closed-source labs are trying to use government intervention to preserve a market position that is eroding far faster than anyone predicted. The central argument of this episode is that Anthropic and OpenAI are engaged in regulatory capture, and that banning open-source models would be a catastrophic self-inflicted wound for the American economy.

The Distillation Disconnect

The core of the debate is "distillation," the process of feeding a model's outputs into the training of another model. The hosts argue that this is a standard industry practice, not IP theft. David Friedberg draws a direct analogy: "At Google in the early days, we would submit millions of search queries to Yahoo and Microsoft search engines to see what the results sets were... It was benchmarking." The speakers make a critical distinction between model weights (the software itself) and model outputs. No one has accused Chinese labs of stealing proprietary weights. They are accused of using outputs to learn, which is exactly what OpenAI and Anthropic argue they have the right to do with the world's data, including from publishers like the New York Times.

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

Never miss an episode of All-In with Chamath, Jason, Sacks & Friedberg

Get every new episode summarized in your inbox — free, ~5 minutes to read.

No spam. Unsubscribe anytime.

What you'll learn

1 (00:01) **Episode Overview** - The hosts introduce the episode's main topics: the debate over banning Chinese open-source AI models, Anthropic's $1.5B copyright settlement, and NYC socialist housing policies.
2 (01:09) **The Kimi K3 Model Sparks a Policy Panic** - China's Moonshot AI releases Kimi K3, an open-source model on par with top US models and 50% cheaper, triggering White House discussions on a potential ban.
3 (04:01) **Sacks: Banning Open Source Would Be a "Tragic Mistake"** - Sacks argues that punishing American developers for using Chinese open-source contributions in the public domain would backfire and hurt US competitiveness.
4 (08:05) **Chamath: Distillation is a Red Herring** - Chamath explains distillation (using model outputs to train new models) is a common practice, and Anthropic itself has "distilled" from publishers.
5 (14:53) **Friedberg: Distillation is Standard Engineering Practice** - Friedberg compares distillation to benchmarking, citing Google's early use of competitor search results to improve its own algorithm.
6 (17:12) **Sacks: The Hypocrisy of "IP for We, Not for Thee"** - Sacks highlights that OpenAI and Anthropic argue they can train on all the world's output (e.g., against the New York Times' wishes), yet condemn Chinese companies for doing the same to them.
7 (20:11) **Friedberg: Three Arguments for Open Source** - Friedberg makes three points: 1) Enforcing a ban on downloaded open-source software is practically impossible. 2) Open source is better for the world, citing the internet's growth via open-source browsers and servers. 3) Open source democratizes AI value, preventing it from accruing only to a few companies.