I’m Crashing Out About Anthropic

I’m Crashing Out About Anthropic

More

Summary

On the day of Anthropic’s IPO, Matthew Berman lays out what they see as the contradictions at the heart of the company. The video traces Anthropic’s origins, founded in 2021 by former OpenAI staff including Dario Amodei and Daniela Amodei who left over concerns about AI safety, and examines how its public messaging compares with its actions.

The host points to several tensions: warning about a white-collar bloodbath while building the technology behind it, the Mythos release that was framed as too dangerous to ship before being released anyway, and a stance against open-weight models and distillation despite training on vast amounts of human-created data. They also argue that lobbying for stricter regulation and GPU controls tends to favor incumbents over startups.

The video recounts the OpenClaw episode, in which Anthropic sent a naming request to creator Peter Steinberger and later restricted subscription use in the tool with what the host calls poor communication. The central argument is that concentrating control of advanced AI in a few hands is a risk Anthropic rarely discusses, and that wider access to AI is a better counterweight. The host notes they know people at Anthropic, receive early model access, and do not think its employees are bad people.


📺 Source: Matthew Berman · Published November 21, 2026
🏷️ Format: Opinion Editorial

1 Item

Channels

1 Item

Companies

1 Item

People