The Download

Adobe Stock

TL;DR: For the first time ever, scientists say they’ve taught AI to create brand-new viruses. The breakthrough could offer a new route for treating infections that antibiotics can’t touch, but you can probably guess the downsides…

What happened: In a new study published in Science yesterday, researchers at Stanford University and the Arc Institute showed that they used AI to write new viral genomes—first by teaching it to recognize the DNA structure of natural viruses, then asking it for blueprints that don’t exist. The team’s AI models, Evo 1 and Evo 2, trained on a massive library of genetic data, then specifically on viruses related to Phi X-174, a natural virus that only infects the E. coli bacteria.

Of the 700,000 potential viruses generated by Evo, nearly 300 were built in the lab, and 16 turned out to be viable. (Some multiplied even faster than Phi X-174.)

Why it matters: A mixture of the AI-designed viruses overcame E. coli strains that had learned to fend off Phi X-174 (which a comparable mix of natural viruses couldn’t do). It’s an early proof of concept that AI could be used to treat drug-resistant bacterial infections. It’s also a turning point for AI use in biology, as models move from analyzing and predicting genomes to writing complete, functional ones.

Voluntary safety: As anyone who saw the Bond movie No Time to Die knows, engineered bioweapon plots don’t end well. And right now, the only thing keeping that scenario fictional is the honor system. The researchers in the study wanted to be “extra careful” about the potential of generating genomes of viruses that could harm humans, and deliberately excluded viruses that could infect humans, animals, and plants from the training data.

Health security scientists at Johns Hopkins, who weren’t involved in the study, commended the team on the precautionary step. As they put it, “The ability to compose viral genomes using generative AI now exists; the governance to safely steer it does not.”

Bottom line: The models are a step toward new treatments, but they’re also a preview of the next AI safety fight. For now, the only guardrail is what a research team decides not to teach its model. —LC

      

Sponsored By Parloa

CX roles are being rewritten by AI. And teams learning to work alongside agentic AI aren’t just adapting to the change—they’re leading the charge. Take charge at Amplitude.

Amplitude is a virtual competition for the CX world’s biggest innovators. In just two hours, you’ll go from a blank brief to a working AI agent in Parloa, no prior experience required. Just bring your big ideas and a little competitive fire.

Folks of all backgrounds and skill levels will get practical experience they can put to work on Monday morning. Bring your team or register solo + get matched with a squad.

Save your spot.

The Zeitbyte

Knit fast and break things

Screenshot via @chamath/X

August isn’t exactly sweater weather—yet Silicon Valley’s current fashion obsession is, inexplicably, venture capitalist Chamath Palihapitiya’s plush white turtleneck.

The trend kicked off sometime last week after Palihapitiya declared on X that wine was his vice of choice. Cut to: an edit of him in the bougie-looking knit, glass of wine in hand. More absurd versions soon followed, spinning into a series of AI-generated memes that got thicker and more enveloping with each iteration—until Palihapitiya was eventually transformed into a giant Dune sandworm.

Elsewhere, the sweater has shown up in edits of Edvard Munch’s The Scream, Michelangelo’s Creation of Adam, and even Ben Affleck’s infamous smoking photo. Other versions are less artistic allusions, more a mockery of Silicon Valley-ites who overuse VC jargon, shoehorning mentions of “founder mode” and “dry powder” into casual conversation. But at the end of the day, this strange woolen saga might simply be another sign that we yearn for the return of high-quality, structurally ambitious knitwear. —WK

Chaos Brewing Meter: ☕☕/5

• A judge in New Mexico ordered Meta to pay an additional $567 million in the child safety case it lost back in March, bringing the total penalty to $942 million.
• OpenAI’s mysterious new device, set for a 2027 release, is reportedly donut-shaped, screenless, and costs more than a basic Apple Watch.
• May the odds be ever in your favor. This quiz tests how fast you can prove that you’re human.
• Amazon is building a data center in Gilroy, California, and, somehow, nobody in the town seemed to notice.
• If you can’t beat them, join them? One AI startup is betting that consumers will want to sell their data directly to AI companies.

^{*A message from our sponsor.}

This time last week...

Readers’ most-clicked story was about this very wrong map of Africa (which has an AI watermark) that was presented by the US government at a global conference. Needless to say, it got blasted on the internet.