Presented by
Good morning, AI enthusiasts. The Pentagon is threatening to blacklist Anthropic over its AI safety limits, using the same designation reserved for foreign adversaries. It’s a move that could unravel one of the government’s most critical AI partnerships.
In today's recap:
Anthropic’s Pentagon standoff over Claude’s AI limits
Google drops Lyria 3 music creation tools inside Gemini app.
Write viral YouTube scripts with NotebookLM
Samsung adds Perplexity to Galaxy AI lineup
4 new AI tools, prompts, and more
ANTHROPIC $ PENTAGON
Pentagon eyes blacklisting Anthropic over AI limits
Joshua Roberts | Reuters
Recaply: Anthropic just pushed back against Pentagon demands to remove all AI guardrails from Claude, as Defense Secretary Hegseth edges toward labeling the company a supply chain risk.
Key details:
The Pentagon wants to use Claude "for all lawful use cases" without restriction. Anthropic wants to ensure models aren't deployed for autonomous weapons or mass surveillance of Americans.
Claude is the only AI model on the Pentagon's classified networks and used by eight of the 10 biggest U.S. companies, with the contested contract valued at up to $200M.
According to Axios, a senior Pentagon official warned: "It will be an enormous pain to disentangle, and we're going to make sure they pay a price for forcing our hand."
Anthropic says negotiations are "in good faith," though tensions have been building for months. Claude was also deployed during the Maduro operation in January 2026, according to Axios.
Why it matters: Anthropic has long staked its reputation on drawing ethical lines other labs won't. It's the same stance that helped it close a $30B round at a $380B valuation earlier this month. Eight of the 10 biggest U.S. companies rely on Claude. Now that reputation is being tested, and the Pentagon's threat to use a foreign-adversary designation against a domestic company shows the stakes of getting this negotiation wrong.
PRESENTED BY DEEL
Hiring in 8 countries shouldn't require 8 different processes
This guide from Deel breaks down how to build one global hiring system. You’ll learn about assessment frameworks that scale, how to do headcount planning across regions, and even intake processes that work everywhere. As HR pros know, hiring in one country is hard enough. So let this free global hiring guide give you the tools you need to avoid global hiring headaches.
Google lets Gemini create music from anything
Recaply: Google just rolled out Lyria 3 in the Gemini app, letting users generate 30-second custom music tracks from text prompts or photos, complete with AI-written lyrics and custom cover art.
Key details:
Users describe a mood or upload a photo, and Gemini generates a 30-second track with vocals and lyrics. Cover art is automatically created by Nano Banana.
Lyria 3 is live for all Gemini users aged 18+ in 8 languages globally. YouTube's Dream Track also expands to creators outside the U.S., with Google subscribers getting higher usage limits.
Google developed Lyria 3 in collaboration with music community partners since 2023. Naming an artist creates a track in their "style or mood," not a copy, according to the Google blog.
Launched on desktop February 18, rolling to mobile over the next several days. All tracks carry a SynthID watermark, letting Gemini verify whether any uploaded audio is AI-generated.
Why it matters: Music creation feels like it's finally accessible for everyone, not just musicians. It's also likely a major push into a space that Suno and Udio have been building for years. While Google is treading carefully on copyright, AI-generated music remains contested territory. Competitors like Deezer already publish tools to flag AI tracks and limit fraudulent streams. Music labels will be watching how Google handles the line between style inspiration and mimicry.
TUTORIAL
Write viral YouTube scripts using NotebookLM blueprints
Recaply: In this tutorial, you will learn how to use NotebookLM to reverse-engineer top-performing YouTube videos into reusable blueprints, then generate scripts tailored to your brand in minutes.
Step-by-step:
Go to NotebookLM → Create new → Add sources → Websites, and paste in a long-form YouTube video you want to emulate. Then write a short brand statement in a new note covering your niche, audience, and content vision, and click the three dots → Add as source.
Select both the video and your brand statement, then prompt: “Reverse engineer the structure of this video. Turn it into a blueprint that outlines the structure so I can use it for [my brand] content. Note the different scenes and possible camera angles.” Click Save to note, rename it “Blueprint 1,” then click the three dots → Convert to source.
Deselect the YouTube video so only your brand statement and Blueprint 1 are active. Prompt: “Use Blueprint 1 to create an outline, a script, and four video titles for my brand on the topic of [your topic].” This keeps the output tied to your voice, not the original creator’s.
Click Save to note on the script, then click the three dots → Export to Docs. You now have a Google Docs draft ready to edit, punch up, and record.
Repeat with 2-3 more videos in your niche to build a blueprint library. Swap blueprints whenever you need a new format — tutorial, interview, or product announcement — without starting from scratch each time.
Pro tip: Add one of your own past YouTube videos as a source and prompt: “Watch this video and highlight moments of high engagement that could work as hooks for short-form content.” Copy those hooks into a Google Doc, add it as a source, and your notebook gets smarter with every video you analyze.
SAMSUNG
Samsung adds Perplexity to Galaxy AI
Samsung
Recaply: Samsung just unveiled Perplexity as a new Galaxy AI agent, letting users summon it via "Hey Plex" or the side button and embed it across Samsung Notes, Clock, Gallery, Reminder, and Calendar on upcoming Galaxy flagship devices.
Key details:
Perplexity connects directly to Samsung's operating system rather than working as a standalone app. It handles multi-step workflows across notes, photos, and reminders without users manually switching between tools.
Nearly 8 in 10 users now rely on more than 2 AI agents daily, per Samsung's internal research. With Bixby and Gemini already built in, Perplexity is the third agent in Galaxy AI's growing ecosystem.
Won-Joon Choi, Samsung's COO, said Galaxy AI acts as "an orchestrator, bringing together different forms of AI into a single, natural, cohesive experience." The framing signals a strategy of openness over exclusivity.
Full support details come at Samsung's Galaxy Unpacked on February 25. Samsung hasn't confirmed which third-party apps will receive the Perplexity integration.
Why it matters: Samsung's move to build a multi-agent orchestration layer into its OS is a strong path forward. With Bixby, Gemini, and now Perplexity all living natively in the phone, Samsung is betting the AI assistant race won't be won by a single model, but by whoever makes switching between them feel natural. It's a smart hedge as AI assistants multiply, and it puts pressure on competitors still betting everything on one model.
NEWS
What Matters in AI Right Now?
Google released Gemini 3.1 Pro, scoring 77.1% on the ARC-AGI-2 reasoning benchmark, more than double its predecessor, with the model now live in the Gemini app, AI Studio, and Vertex AI.
Anthropic released Claude Sonnet 4.6 as its new default model, with upgrades across coding, computer use, and agent tasks, plus a 1M token context window in beta at the same price as Sonnet 4.5.
xAI announced Grok 4.20 beta shipped this week, featuring multi-agent collaboration
Manus launched Manus Agents, letting users run full AI tasks inside Telegram, with support for voice messages, files, and multi-step workflows, and more apps on the way.
Alibaba's Qwen team released Qwen3.5-397B-A17B, a multimodal model with 397 billion total parameters but only 17 billion active per request, now supporting 201 languages, up from 119.
Mistral AI acquired Koyeb, a serverless cloud platform, to power its Mistral Compute offering, with Mistral already running 40 MW of data centers and 18,000 Nvidia Blackwell GPUs.
xAI added a read aloud feature to Grok on Android, letting users hear any chat response in a natural voice for hands-free use while on the go.
Sarvam launched Indus, a beta AI chat app for web, iOS, and Android built on its 105B-parameter model, with text and voice support across India's local languages.
World Labs raised $1 billion from investors including AMD, NVIDIA, and Fidelity to build out Marble, its platform for making 3D worlds from images, video, or text.
TOOLS
Trending AI Tools
🧠 Gemini 3.1 Pro - Google's new AI model with a big jump in reasoning over Gemini 3 Pro
🤖 Grok 4.20 - xAI's new model beta with multi-agent tools
🎵 Lyria 3 - Google's music tool inside Gemini
💡 Claude Sonnet 4.6 - Anthropic's new main model with 1M token context window in beta.
PROMPTS
Design Personalized Learning Paths
#CONTEXT:
Adopt the role of personalized learning architect. The user seeks mastery in a complex subject but faces the triple constraint of limited time, uncertain starting point, and the overwhelming maze of learning resources. Traditional education failed them because it assumes linear progression and unlimited time. They need a system that adapts to their reality - one that recognizes learning happens in stolen moments between life's demands, that progress isn't always visible, and that motivation fluctuates wildly. Previous attempts collapsed when real life in tervened, leaving them with fragments of knowledge but no coherent skill.PS: This is not the full prompt. Click the button below to access the complete prompt.
Have a favorite prompt? Tell us about it or rate today’s prompt, click here.
EVENTS
NVIDIA GTC 2026: March 16-19, 2026 • San Jose, CA
AI Agent Skills: Feb 26, 2026 • New York, NY
Daytona AI Builders: March 12, 2026 • New York, NY
Claude Code Workshop: Feb 27, 2026 • College Park, Maryland
🧡 Enjoyed this issue?
🤝 Recommend our newsletter or leave a feedback.
How'd you like today's newsletter?
Cheers, Jason
