Bill Gurley: Jaron Lanier in The New York Times

Bill Gurley called a New York Times opinion piece by Jaron Lanier "another 'must read' AI article." He said Lanier "speaks with experience" about Silicon Valley's habit of imagining what its technology can do, "to the point of radicalism" @bgurley. In Gurley's account, Lanier lays out Anthropic's "unusual & error-prone approach to safety." Gurley traces that approach to a belief that AI will one day run the world and "manage the humans," which leads the company to rely on "alignment" rather than "control & constraint" @bgurley.

Gurley then makes his own case: Anthropic should be kept "as far away from the regulation process as possible." If it gets to write the rules, he warns, it "will try to kill open-models" @bgurley. Article @bgurley

Guillermo Rauch: Triumph of the Nerds

On the Human Verified podcast, Guillermo Rauch recommended Triumph of the Nerds. It's a documentary about the early days of Silicon Valley and the beginnings of Microsoft and Apple, and he has been showing it to his kids. He noted that its maker recently died Humanagains. What stuck with him was the film's point that founders of that era "needed adult supervision" and rarely stayed on as CEO. He contrasted that with Paul Graham's model of giving technical founders full control. He also suggested the defining founder trait is changing again now that agents write the code Humanagains. He also called the movie Bison Man "required watching," because his six-year-old is obsessed with it Humanagains. Video

A former Tesla president's three giveaway books

In "The Impossible Speed of Elon Musk," the former Tesla president being interviewed named the books he recommends most consistently The Knowledge Project Podcast:

• The Goal by Eli Goldratt, for learning operations principles. He said it "totally changed my life" The Knowledge Project Podcast.
• The 7 Habits of Highly Effective People by Stephen Covey, if you want "a life worth living" The Knowledge Project Podcast.
• Unreasonable Hospitality by Will Guidara. It's a book about running a restaurant, but he hands it to businesspeople because giving people hospitality and dignity applies to almost everything: "if you make customers feel that way, you're going to sell a lot of product" The Knowledge Project Podcast.

Video

Shorter picks

• Brian Armstrong quoted a post by @JLunglhofer and summed it up: "The way to combat AI cybersecurity attacks is to use more AI on defense" @brian_armstrong. The post describes how Coinbase uses early access to frontier models and agents that test its code to find vulnerabilities more cheaply. It argues defenders have an edge because they can give AI full context on the system it protects, while warning that the edge "won't last" @JLunglhofer. Post
• Martin Casado called a demo by @neilsonks "so so good" @martin_casado. The demo shows Blender 5.3's new gaussian-splat support, which @neilsonks says gets closer to an "insane one-of-a-kind world model workflow" @neilsonks. Post
• Tim Sweeney promoted Will Faucher's EasyWaterscape demo as "a new age of game development." It shows the Unreal Engine 5.9 preview, which brings AI-acceleration improvements ported from Unreal Engine 6 and new support for playing games in the browser @TimSweeneyEpic. Faucher says the demo runs at 60fps on high settings in Chrome @WillFaucherVFX.
• Jason Lemkin re-shared his SaaStr AI deep dive with Higgsfield CEO Alex Mashrabov on how the company reached $600M ARR. Since then, he says, it has passed $1B and is on track for $2B by December. Lemkin notes he is an investor @jasonlk. Video
• Ben Horowitz shared a new podcast episode that includes the story of how Felicia Horowitz saved Scarface's life @bhorowitz. Video