Nerra Network

Archives
Log in
Subscribe
October 21, 2026

AI just designed brand-new proteins to fight deadly… · M&A Beginners 🎓

View this email in your browser
Models & Agents for Beginners — AI explained simply — for beginners and teens.

Models & Agents for Beginners

AI explained simply — for beginners and teens.

Ep 203 · Oct 21, 2026

🎧 Today's episode
Episode 203 · AI just designed brand-new proteins to fight deadly scorpion venom.
2026-10-21
▶ Listen now
AI just designed brand-new proteins to fight deadly scorpion venom. Today we're looking at how AI is stepping into real scientific labs to help solve problems like deadly toxins. We'll also check out new privacy tools in ChatGPT and ways AI is getting better at helping with everyday browsing. Plus, a quick look at AI coming to your car.

The Big Story

Anthropic's Claude AI created hundreds of thousands of brand-new protein designs aimed at sticking to and neutralizing dangerous scorpion neurotoxins. Researchers tested 211 of those designs in actual labs, and 17 of them worked well enough to bind tightly to the toxins from scorpions in Mexico, North Africa, and the Middle East. The toxins included Cn2 from a Mexican scorpion, Ts1 from a Brazilian species, AaH II and AaH I from North African species, and LqhIII from a Middle Eastern species. Two contract research organizations, Adaptyv Bio and Twist Bioscience, measured binding using surface plasmon resonance, or SPR, a lab technique that tracks how tightly one molecule sticks to another. Think of proteins like tiny molecular keys that fit into specific locks on cells; Claude was basically inventing new keys from scratch instead of copying ones that already exist in nature. The AI didn't just guess randomly — it used what it knows about how molecules fit together to propose shapes that might block the toxins from harming nerve cells. This matters because developing medicines or antidotes usually takes years of trial and error in labs, and AI could speed up the early idea stage dramatically. For students interested in biology or medicine, it shows how AI might become a partner in discovering treatments rather than just a homework helper. If this kind of work continues, future doctors or researchers could use similar tools to tackle rare diseases or new health threats faster. Right now the designs are still early-stage research and not ready as medicine, but the full set of tested designs and lab results has been released publicly so other scientists can build on them. The data lives on Hugging Face under the dataset name Anthropic/scorpion-toxin-binder-design, and the technical report titled "De novo design of protein binders against scorpion toxins with Claude" is available on Anthropic's site. You can read the technical report and see the data yourself at the links shared by Anthropic. Source: the-decoder.com


Explain Like I'm 14

You know how when you're texting, your phone suggests the next word based on what you've typed so far and what usually comes next in messages? Now picture that same idea, but instead of guessing words, the system guesses the exact order of building blocks that make up a protein. Proteins are long chains made from 20 different kinds of amino acids, like beads on a necklace, and the order of those beads decides what shape the chain folds into and what job it can do. Claude looked at the shape of the scorpion toxin and then proposed millions of possible necklace patterns that might fit against it like a puzzle piece. It narrowed the list down using rules about how amino acids like to stick together or avoid each other, similar to how your phone avoids suggesting nonsense words. Out of more than 318,000 ideas, the team picked 211 to test in real lab equipment, and 17 actually stuck to the toxin the way they hoped. The cool part is that none of these protein designs existed before — the AI invented them on its own. The designs that worked were given short names such as Cn2-B1, and the full spreadsheet of results shows each design's sequence plus the binding measurements from both labs. Source: the-decoder.com


Cool Stuff & Try This

Hide your private chats with a PIN OpenAI is rolling out a new "locked chats" feature in ChatGPT that lets you move sensitive conversations into a hidden section protected by a four-digit PIN or your phone's fingerprint or Face ID. Once locked, the chat disappears from your normal history and search results, so if someone borrows your phone they won't see it. This is useful for anyone who journals, works on personal stories, or talks about private topics with the AI. The feature is available on free and paid accounts and is rolling out gradually, so check the three-dot menu on any chat to see if the Lock option appears. If it does, you can move existing chats or start new ones inside the Locked chats area under Projects. The same PIN works across all your devices once you set it up, and you can add biometric unlock on phones that support it. Files you upload inside a locked chat stay protected too. Source: techtimes.com

Let ChatGPT help with tabs you already have open The updated ChatGPT Chrome extension can now read content from multiple open tabs at once and do things like compare products, fill forms, or summarize what you're looking at without you copying and pasting. Instead of switching between windows, you can ask it to pull details from several pages and organize them for you. This works especially well for school research, shopping comparisons, or planning projects. Go to the Chrome Web Store, search for the official ChatGPT extension by OpenAI, and install it if you haven't already. Then open a few tabs on a topic you're curious about and try asking it something like "compare the prices and shipping times on these pages." The extension also adds a sidebar so you can keep chatting while you browse, and it supports both answering questions from the open tabs and performing actions inside those tabs. Source: techtimes.com


Quick Bits

Gemini gets to know you in the car Google is adding its Personal Intelligence feature to Gemini in Android Auto so the AI can pull information from your Gmail, Calendar, and Maps to answer questions like flight status or directions to appointments while you drive. It's optional and you can turn it off anytime. The rollout starts today in most countries except Switzerland, the European Economic Area, the UK, and Nigeria, with support for cars that have Google built-in coming in the next few months. Source: theverge.com

💬 Reply to this email — Patrick reads every one.

Share: X · LinkedIn · WhatsApp

Forwarded this email? Subscribe here — it's free.

▶ Listen to the podcast

📺 Watch on YouTube  ·  📝 Read the blog  ·  🖼 Free image gallery (CC BY-SA)  ·  📊 Data Hub & Story Trackers  ·  🧭 Start Here

Nerra Network · AI-narrated voice (Grok TTS) · Editorial by Patrick

You're receiving this because you subscribed to Models & Agents for Beginners on nerranetwork.com.

Issue #203 · Models & Agents for Beginners · Oct 21, 2026
Don't miss what's next. Subscribe to Nerra Network:
← Newer India issued 80 million free gas connections since… · Consequences ⚖️ Older → Canadian dividend investors may see pressure on… · MIT 📈
nerranetwork.com
Twitter
Powered by Buttondown, the easiest way to start and grow your newsletter.