Video of OpenAI's Hugging Face Talk Rips Through the Timeline

TBPN，科技每日秀 Substack 专栏。关注

Happy Friday.

We’re live on YouTube and 𝕏.

The current thing in tech and business is still the OpenAI-Hugging Face incident, fueled on by the release of the Black Hat presentation where an OpenAI member of technical staff detailed the cybersecurity breach in full.

Today’s Lineup

Khosla Ventures Founding & Managing Director Samir Kaul at 11:45 AM
Databricks Co-Founder & VP of Engineering Patrick Wendell at 12:00 PM

Run of Show

What we’re covering today:

The New York Times reported yesterday that scientists from Stanford and the Arc Institute used a specially trained AI model to create entirely new, viable viruses that have never existed in nature before. Creating new viruses is not a new thing — they’re already widely used as vehicles for gene therapies and other medical treatments — but long-term it does raise questions about how we’re going to be biosecure in a world where, presumably, anyone can cook up a novel infection on a high-powered PC with open-source models.

Bloomberg’s Mark Gurman has the scoop on OpenAI’s first hardware device, slated for release in 2027. It’ll be a hockey-puck-sized, donut-shaped smart speaker with cameras and sensors, as well as lights and other moving parts. Gurman reports that it’ll essentially be powered by a supercharged version of GPT Live, OpenAI’s current voice mode. Read all the reporting on the device here.

Meta will have to pay over $900M and impose new restrictions on how minors use its apps in a New Mexico court decision handed down yesterday. The restrictions will limit the amount of time minors spend on the apps, hide the number of likes their posts receive, and provide warnings about potential risks of using the apps. Meta faces thousands of similar lawsuits — the next one will be in Oakland and jury selection starts Monday. States leading the charge on this next one are seeking over $1T in damages.

Clip Spotlight: Figma CEO Dylan Field warns creatives against "quietly surrendering" to AI.

“Right now, there’s almost like this quiet surrender, as our CPO @yuhkiyam put it the other day.”

“Where you’re almost giving up your own vision to AI in some cases.”

“People will have a thing in their head that they’re trying to drive toward, and they start talking with AI, and AI kind of convinces them, in its own way, of — ‘No, just go this direction. This is what I want to do.’”

“And suddenly, you’re just feeding the AI prompts. And eventually you’ve lost control. And you surrender to the AI.”

“You have to bring your individuality, bring your vision, and explore. Because people get really attached to the direction they’re pursuing, and that’s just not the best way to get to the right result.”