Presented by
Good morning, AI enthusiasts. The U.S. government ordered Anthropic to pull Fable 5 and Mythos 5 offline on a Friday afternoon phone call, with no prior warning about any national security threat. To comply, Anthropic disabled both models for every customer.
The White House is now talking about building a formal AI security framework, not resolving the ban. That means the impasse isn't ending anytime soon.
In today's recap:
U.S. and Anthropic deadlocked on AI export controls
Half of U.S. adults now use AI chatbots
Share a live session page with Claude Code Artifacts
Domain expertise beats coding skills in Claude Code
4 new AI tools, prompts, and more
ANTHROPIC
U.S. and Anthropic locked in AI export control impasse
Getty Images
Recaply: Anthropic remained locked in an impasse with the U.S. government over export rules that took Fable 5 and Mythos 5 offline, with White House talks now shifting from ending the ban to building a formal AI security framework.
Key details:
A government directive told Anthropic to cut off Fable 5 and Mythos 5 for all foreign nationals, inside or outside the U.S. To comply, Anthropic shut down both models for every customer at once.
About 200 companies in Project Glasswing, a cybersecurity partner program, kept Mythos Preview access despite the ban, including Cisco, AWS, and JPMorgan Chase. The DOD had flagged Anthropic as a supply chain risk in March, and that suit is still open.
Amazon CEO Andy Jassy was among the tech leaders who told Trump officials the models posed a security risk, a White House official confirmed to CNBC.
Anthropic got the government's call at 1 p.m. ET Friday and a formal letter by 5:30 p.m. The company had received no prior warning about any national security threat, a source close to the matter said.
Why it matters: Anthropic got approval to ship Fable 5 and Mythos 5, then received a shutdown order four days later with no warning. That gap matters more than the ban itself. The White House is now building an AI security framework from scratch, which means the U.S. has no stable process yet for vetting frontier models. Every major lab is watching how this ends.
PRESENTED BY STORE CLAW
One AI. Every Tool Your Store Actually Needs.
Most e-commerce sellers are paying for 6 to 8 separate tools that don't talk to each other — and spending hundreds of dollars a month just to keep up. StoreClaw replaces your entire stack with one autonomous AI engine that monitors competitors, optimizes listings, automates marketing, and tracks real profit across Shopify, Amazon, and beyond.
It doesn't wait for you to ask. It runs 24/7 in the background, so you wake up to a full dashboard instead of a list of things you forgot to check.
Connect your store, and StoreClaw gets to work — no prompts, no complex setup, no six-app stack.
Free to start. No credit card required.
PEW RESEARCH
Half of U.S. adults now use AI chatbots, but views on pace tilt negative
Pew Research Center
Recaply: Pew Research Center just released its 2026 Americans and AI survey of 5,119 U.S. adults, finding that about half now use AI chatbots. That's up from a third in 2024, but views on AI's overall impact tilt negative across all age groups.
Key details:
Survey ran Feb. 17-23, 2026, covering chatbot usage, AI-enabled smart devices, AI search summaries, and how Americans feel about AI's societal impact across the full U.S. adult population.
Chatbot use rose from about 33% of U.S. adults in 2024 to roughly 50% in 2026. The 5,119-person sample came from Pew's American Trends Panel.
Views on how fast AI is moving tilt negative even among younger adults, per the report. That complicates the idea that digital-native generations are comfortable with AI's pace.
The survey covers seven chapters, including a gender gap in AI use, racial and ethnic differences in adoption, and an appendix of detailed tables available for download.
Why it matters: Half of U.S. adults now use AI chatbots. But the bigger signal is who's not, and why. Pew's dataset is the first major 2026 snapshot of mainstream adoption and sentiment. The negative skew on AI's pace isn't coming from technophobes. It's coming from the general population, including younger adults. For anyone building consumer AI products, that's a perception gap that won't close by shipping faster.
GUIDES
Share a live session page with Claude Code Artifacts
Recaply: In this tutorial, you will learn how to create and share an Artifact in Claude Code, turning your session's work into a live, interactive page your team can view at a private URL without copying terminal output.
Step-by-step:
Confirm your setup. Sign in to Claude Code on a Team or Enterprise plan using
/login, and have an active coding session open with your codebase loaded. Artifacts require an org-level login, not a personal account.Run your task as normal, then ask Claude to build an Artifact when the output suits a visual format. For a PR review, try: "Make an artifact that walks through this pull request with the diff annotated inline."
When Claude asks for permission to publish, select Yes. Claude writes the page as an HTML file in your project, publishes it to a private URL on claude.ai, and opens it in your browser. Press
Ctrl+]at any time to reopen the latest Artifact from the terminal.Open the Share control in the page header to grant access to specific teammates or everyone in your organization. Anyone you share with must sign in to claude.ai as a member of the same org to view it.
Ask Claude to keep the page updated as the session continues. Try: "Add a per-region breakdown and republish." Each publish hits the same URL, and anyone with the page open sees the update in place. Use the Share control to choose which version they see.
Pro tip: Share the Artifact link before the session is done. Since each publish replaces the same URL, your reviewer always sees the latest version without needing a new link.
TOGETHER WITH MARKETING AGAINST THE GRAIN
50 Excel Hacks That Make Spreadsheets Work For You
Most Excel users waste hours on tasks that take experts minutes. The difference is not talent. It is knowing the right shortcuts.
These 50 Excel hacks from Kenji Explains cover the functions, formulas, and workflows that make spreadsheets work for you instead of against you.
Each one comes with step-by-step instructions and ready-to-use templates. Subscribe to Marketing Against the Grain and get all 50 free.
Here's what you'll get:
Time-saving shortcuts that eliminate the most common formula frustrations
Interface essentials most users never discover on their own
Game-changing functions for data analysis, automation, and visualization
Guided practice template so you can apply every hack immediately
The people who are fast in Excel are not smarter. They just stopped fighting the tool.
Get 50 Excel Hacks free when you subscribe to Marketing Against the Grain today.
AI RESEARCH
Domain expertise, not coding skills, unlocks Claude Code's full output
Anthropic
Recaply: Anthropic just published a study of about 400,000 Claude Code sessions from 235,000 users, finding that domain know-how, not coding skill, drives how much the agent does per instruction and whether sessions end in success.
Key details:
In a typical session, people make about 70% of the planning calls while Claude handles about 80% of the execution. Each user prompt triggers an average of 10 Claude actions and 2,400 words of output.
Expert users trigger 12 Claude actions and 3,200 words of output per prompt. Novice users trigger just 5 actions and 600 words, a gap that holds across every type of work and task value band.
Domain know-how beats coding skill. A lawyer who knows which contract clauses a script must catch succeeds at nearly the same rate as a software engineer, according to the study.
Over seven months from October 2025 to April 2026, debugging sessions fell by nearly half. The typical task value rose about 25% as usage shifted toward end-to-end agentic work.
Why it matters: If domain know-how amplifies AI coding performance more than coding skills do, the labor market math shifts. Lawyers who know what contracts need and accountants who know what a reconciliation script must catch get more from AI coding agents than generalist coders who don't know the domain. That's not a threat to domain experts. It's a threat to the implementation-only layer of software work.
TOOLS
Trending AI Tools
💻 Claude Code Artifacts - Anthropic's beta feature that turns Claude Code sessions into live interactive pages, shareable at a private org link, for Team and Enterprise plans
⚙️ Codex Record & Replay - OpenAI's new Codex plugin that records a workflow demo once and turns it into a reusable, editable skill
🧠 Perplexity Brain - A continuously learning memory system for Perplexity Computer that improves answer correctness by 25% and recall by 16% per context-dependent task
🔧 Zero-Touch OAuth for MCP - MCP's Enterprise-Managed Authorization extension that replaces per-user OAuth prompts with single org-level sign-on for all connected servers
NEWS
What Matters in AI Right Now?
General Intuition is in talks to raise $300M at a roughly $2B valuation, eight months after spinning out of Medal with a $134M seed, training embodied AI on Medal's dataset of 2 billion gaming videos per year. Backers include Jeff Bezos, Eric Schmidt, Khosla Ventures, and General Catalyst.
Claude just rolled out Artifacts in Claude Code, turning sessions into live interactive pages, like PR walkthroughs or project dashboards, shareable via private link inside your organization. Artifacts refresh automatically as the session continues and are available in beta on Team and Enterprise plans.
Perplexity just introduced Brain in Computer, a memory system that feeds Perplexity Computer a full context graph before each task instead of starting from scratch. On context-dependent tasks, Brain improves answer correctness by 25% and recall by 16%, and runs 13% cheaper per task. Available in research preview for Max subscribers.
OpenAI just launched Record & Replay for Codex, letting you show the agent a recurring workflow once and have it turn that demo into an inspectable, editable skill it can run on demand. Rolling out to select markets with broader availability to come.
Google just introduced Ask Ad Manager, an AI agent built with Gemini that lets publishers troubleshoot ad issues, generate custom performance reports, and navigate platform settings through natural conversation. The beta launches this month with more features rolling out through the year.
🧡 Enjoyed this issue?
🤝 Recommend our newsletter or leave a feedback.
How'd you like today's newsletter?
Cheers, Jason
