Tao Lin

Things Tao Says on Podcasts

I'm the author of ten books, including Leave Society, Trip, and Taipei. I live in Hawaii.

Where to Find Them

Tao Lin writes The Paris Review , writes early girl , writes The Metropolitan Review and writes The Republic of Letters . They have also been a guest on AI Explained Official Podcast . They also write for Tao Lin on Substack.

Recently: “Anthropic IPOs, GPT 6.2 Astra Arrives, But There’s a Bigger Story” on AI Explained Official Podcast (November 2026); “Taiwan English” on The Paris Review (June 2026); “A Car Hit My Bike” on The Metropolitan Review (February 2026); “An Interview With Tao Lin” on The Republic of Letters (November 2025); “My College Diary” on The Paris Review (August 2025); “Check Please! #7 - Malaise, Raw Milk & a Chat with Tao Lin (Sept. 10, 2024)” on early girl (September 2024).

What They Said

The closing clip on AI Explained, introduced as Anthropic employee Tao Lin. He questions the reasoning of people who join Anthropic to build capabilities because they believe a safer lab should win, and calls it a very risky bet.

AI Explained Official Podcast · 2026-11-19 Permalink → Listen →
AI Explained Official Podcast Around 25:06 into the episode
Speaker 2

And I believe that the company which gets the risks right will also be the company that has the license to operate, that society trusts and society believes can build the benefits in the right way.

Speaker 1

But then here's Tao Lin, a humble Anthropic employee. He thinks it would be, quote, insane for humanity to rely on helping Anthropic win just because it is marginally safer.

Speaker 4

I think many people work on Capabilities Anthropic because they believe that empowering a safer actor is better. I think that this is very far from sufficient. A lot of people at Anthropic are trying to help Anthropic win the AI race. I think a lot of people at Anthropic. Are trying to help empower companies that will be safer and that will prioritize ensuring that they can align and control their AIs before building them or before deploying them. And that this is enough reason to work on capabilities in Anthropic. I think that this is a very risky bet. And I think that it would be very insane for humanity to rely on that strategy of trying to help Anthropic win because Anthropic will be marginally safer than other labs.

Speaker 1

Let me know what you think about any of this or the models just released. Thank you so much for watching and have a wonderful day.

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

Collections They Appear In