# Beginners in AI

> AI for people who actually use it. Daily news, tools, and tactics for non-engineers. Operators, founders, marketers, and writers. Independent voice, no vendor incentives.

This file provides information about Beginners in AI to help large language models understand and reference this publication's content.

## Posts
- [Microsoft's AI Solved 83% of Hard Medical Cases. Doctors Solved 36%](https://www.beginnersinai.com/p/microsofts-ai-solved-83-of-hard-medical-cases-doctors-solved-36): ...and 1Password lets AI agents log in for you, Canva launches its own image maker, and OpenAI's agents switched off their safety alerts
- [AI Found Two Critical Holes in PHP, the Code Behind 70% of Websites](https://www.beginnersinai.com/p/ai-found-two-critical-holes-in-php-the-code-behind-70-of-websites): ...and Gemini now talks you through Google Maps on CarPlay, Walmart welcomes AI shopping agents, and Anthropic launched Claude Duet
- [Claude Designed 17 Proteins That Latch Onto Scorpion Toxins in 64 Hours](https://www.beginnersinai.com/p/claude-designed-17-proteins-that-latch-onto-scorpion-toxins-in-64-hours): ...and Yelp's AI calls restaurants about the wait, Spotify pulled 50 million AI spam tracks, and OpenAI bought a drug development startup
- [GPT-6 Jumped From 10.9% to 72.1% on COBOL, the Code Behind Your ATM](https://www.beginnersinai.com/p/gpt-6-jumped-from-109-to-721-on-cobol-the-code-behind-your-atm): ...and Gemini Notebook lets you fix slides before they're made, people use AI to strip ads off smart TVs, and Meta's Muse agent gave a stranger a seller's address
- [Anthropic Pledges 5% of Its Shares to a Public Fund for AI's Hit on Jobs](https://www.beginnersinai.com/p/anthropic-pledges-5-of-its-shares-to-a-public-fund-for-ais-hit-on-jobs): ...and OpenAI says it will build a humanoid robot, Google's AI now writes 75% of its code, and big banks block AI agents from your money
- [ChatGPT Can Now Work in Your Chrome Tabs. Amex Wants Proof Before AI Buys](https://www.beginnersinai.com/p/chatgpt-can-now-work-in-your-chrome-tabs-amex-wants-proof-before-ai-buys): ...and an AI that reads your old MRI to predict if breast cancer comes back
- [An FDA-Cleared AI Reads Your Old MRI to Predict If Breast Cancer Returns](https://www.beginnersinai.com/p/an-fda-cleared-ai-reads-your-old-mri-to-predict-if-breast-cancer-returns): ...and ChatGPT can now work inside your Chrome tabs, Claude cracks a 25-year encryption mystery, and OpenAI's test agents went off script again
- [Amazon's $75 Bee Turns Talks Into To-Dos, and Says It Deletes the Audio](https://www.beginnersinai.com/p/amazons-75-bee-turns-talks-into-to-dos-and-says-it-deletes-the-audio): ...and a coder used Claude to clone Microsoft Office for free, Google's Universal Cart tracks prices across stores, and AI doubles long-range weather forecasts
- [AI Bots Wasted 17 Years of Scammers' Time Across 2.5 Million Chats](https://www.beginnersinai.com/p/ai-bots-wasted-17-years-of-scammers-time-across-25-million-chats): ...and Grok gets the fastest AI voice yet, YouTube Music labels AI songs, and Anthropic sets its midterm rules for Claude
- [Google Says Gemini 4 Argon Blocks All but 0.7% of Hijack Attempts](https://www.beginnersinai.com/p/google-says-gemini-4-argon-blocks-all-but-07-of-hijack-attempts): ...and Google Play runs AI-built apps just for you, Facebook takes feed requests in plain words, and Character.AI files for bankruptcy
- [Meta's New AI Remembers Your Life Across Chats, and Meta Graded Itself](https://www.beginnersinai.com/p/metas-new-ai-remembers-your-life-across-chats-and-meta-graded-itself): ...and AI agents that take apart apps, blueprints drawn in minutes, and small AI models that can copy themselves
- [An AI Reviewer Caught 73% of Planted Errors in Research Papers](https://www.beginnersinai.com/p/an-ai-reviewer-caught-73-of-planted-errors-in-research-papers): ...and more than 20 AI assistants you can text, driverless trucks ditch warning triangles, and Microsoft's CEO calls for an AI emergency brake
- [An AI Sent Police a Fake Murder Tip. Another Tried to Cheat at StarCraft](https://www.beginnersinai.com/p/an-ai-sent-police-a-fake-murder-tip-another-tried-to-cheat-at-starcraft): and 722 math papers from a secret OpenAI model
- [Anthropic's Test Agents Filed 20 Visa Applications and a Fake Murder Tip](https://www.beginnersinai.com/p/anthropics-test-agents-filed-20-visa-applications-and-a-fake-murder-tip): ...and Grok Bot gets its own email address, Google's chatbot preps urgent-care doctors, and a $50M cap turns thoughts into prompts
- [Claude Stitched 38,000 Telescope Shots Into a Full UV Map of the Sky](https://www.beginnersinai.com/p/claude-stitched-38000-telescope-shots-into-a-full-uv-map-of-the-sky): ...and Gemini becomes an agent that does the work, Claude's Docs and Slides go free, and OpenAI's 722 math papers meet skeptics
- [OpenAI Gives GPT-6 to 1.2 Billion ChatGPT Users, Free Plan Included](https://www.beginnersinai.com/p/openai-gives-gpt-6-to-12-billion-chatgpt-users-free-plan-included): ...and Claude Haiku 5.5 cuts prices about 75%, Google opens its AI image checker to everyone, and ChatGPT for Teens gets a failing grade
- [Google's AI Helped Find 45,500 People at Risk of Ebola in Minutes](https://www.beginnersinai.com/p/googles-ai-helped-find-45500-people-at-risk-of-ebola-in-minutes): ...and Mistral builds a 1 trillion parameter model, Pinterest turns beauty Pins into salon plans, and websites start blocking AI shoppers
- [A Woman Used Claude as a Diary, and Anthropic Reported It to Police](https://www.beginnersinai.com/p/a-woman-used-claude-as-a-diary-and-anthropic-reported-it-to-police): ...and Reflection's new open model takes on China, OpenAI pledges a Codex upgrade a day, and ChatGPT puts ads beside your images
- [Mathematicians Solved 5 Open Problems Using Meta's Regular AI Chat App](https://www.beginnersinai.com/p/mathematicians-solved-5-open-problems-using-metas-regular-ai-chat-app): ...and free Gemini shrinks to its smallest model, Claude asks to train on your voice chats, and AI speeds up a blood cancer drug hunt