Sponsored by |
 |
|
|
|
| |
| |
Good morning, AI enthusiast. Stanford and Arc Institute scientists just proved something that sounded like science fiction a year ago: AI can write a working virus genome from scratch, with no human editing required. |
The 16 synthetic bacteriophages beat drug-resistant E. coli in the lab, offering a real answer to the antibiotic-resistance crisis. But the same generative approach that built a cure could just as easily design a threat - so how fast can biosecurity rules catch up to what the models can already do? |
Today in AI Brief: |
Stanford's AI writes 16 new viruses from scratch Meta launches Muse Code, an autonomous coding agent Google's Ask Maps now orders your food |
|
|
|
| |
|
How Jennifer Aniston's LolaVie brand grew sales 40% with CTV ads |
|
The DTC beauty category is crowded. To break through, Jennifer Aniston's brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category. |
Learn more |
| |
| |
Stanford's AI Designs 16 Brand-New Viruses |
In Brief: Stanford and Arc Institute researchers used AI models Evo 1 and Evo 2 to design 16 functional bacteriophages that don't exist anywhere in nature, according to a study published in the journal Science. The AI-generated viruses target bacteria, not humans, and a cocktail of them wiped out drug-resistant E. coli strains that natural phages couldn't touch. |
The Details: |
Evo 1 and Evo 2 generated the genomes end-to-end in a single pass, trained on more than 2 million existing bacteriophage genomes, according to the peer-reviewed study. Researchers synthesized roughly 300 candidate genomes and found 16 viable, functional viruses, some replicating faster than anything found in nature. The team deliberately excluded any human, animal, or plant pathogen data from training, limiting the models to bacteria-only phages. |
Take Away: |
Antibiotic-resistant infections kill an estimated 35,000 people a year in the U.S., and phage cocktails that keep pace with mutating bacteria could become a genuine new weapon against them. Johns Hopkins biosecurity researchers who co-published alongside the study warn the same generative technique could just as easily draft dangerous pathogens, and the oversight to prevent that doesn't exist yet. |
|
|
|
| |
|
| |
| |
Meta Launches Muse Code, an Autonomous Coding Agent |
In Brief: Meta released Muse Code in beta, a terminal-based coding agent powered by its new Muse Spark 1.2 model, built to plan, write, and verify changes across large codebases with less human oversight. |
The Details: |
Muse Code runs persistent background agents that stay alive across a session instead of spinning up fresh per task, cutting latency on multi-step jobs and letting agents resume after a crash. Muse Spark 1.2 was co-trained on long-horizon coding tasks, including whole-repository generation, and ships with built-in /plan, /grill, and /goal skills for approval-gated workflows. In one test, the model iterated for 24 hours across 1,000+ tool calls to optimize GPU kernels for Nvidia Hopper chips, install it via Meta's developer portal. |
Take Away: |
Persistent agents that survive crashes and verify their own work move coding assistants closer to genuinely autonomous engineers rather than glorified autocomplete. Developers can try Muse Code today instead of waiting for a wider release. |
|
|
|
| |
|
|
The first way to trade directly inside Claude and ChatGPT |
|
For decades, the most powerful intelligence lived behind the closed doors of quant firms - billion-dollar funds whose algorithms quietly out-traded everyone else. |
That era just ended. |
Co-Invest by Liquid is the first way to trade directly inside Claude and ChatGPT. Ask your AI to analyze a market, stress-test an idea, or build a position sized to your comfort level, then execute, right there in the conversation. No jargon. No twelve-screen terminal. No guesswork. |
It's built for people who want to invest smarter, not gamble harder. You set the risk tolerance. The AI does the heavy lifting. You approve every trade. |
The institutions made the game, Co-Invest gives you a way to beat them. |
Trade with Co-Invest Today |
| |
| |
Google's Ask Maps Now Orders Your Food |
In Brief: Google rolled out new agentic features for Ask Maps, letting users type a request like "order me pad kee mao for pickup on my way home" and have the assistant find an open restaurant, apply dietary preferences, and fill the cart automatically. |
The Details: |
The feature integrates with Square, Toast, and soon Uber Eats, first rolling out to U.S. users before expanding to more countries, TechCrunch reported. Ask Maps can now compare hotel pricing in real time and surface nearby ticketed events alongside directions. Users can optionally connect Gmail, off by default, so Ask Maps can tailor suggestions around upcoming travel reservations and plans. |
Take Away: |
Folding ordering, booking, and trip context directly into a maps app turns navigation into a general-purpose agent that acts on a user's behalf, not just a tool that shows the way. It's a preview of how quickly agentic features are moving from chatbots into everyday utility apps. |
|
|
|
| |
|
| |
| |
Everything else in AI |
Anthropic overhauled Fable 5's biology safeguards, cutting false-positive refusals by roughly 85% while keeping dual-use virology and toxicology requests blocked. |
OpenAI asked a federal judge to dismiss Apple's trade-secret lawsuit, arguing the claims are meritless and noting roughly 400 former Apple employees now work at OpenAI. |
Alibaba's Qwen priced its new Wan3.0-Video model at $0.05 to $0.20 per second, generating up to 30-second clips from text, image, audio, or video prompts. |
Amazon and Microsoft backed Agent Plugins 1.0, a new open packaging standard for AI agent skills and MCP servers, alongside Cursor, OpenAI, and Vercel. |
|
|
|
| |
|
|
|
