In partnership with
|
THIS WEEK
|
The OpenClaw Story Has Three Acts and None of Them Are Simple
If you've been following tech news this week, you've heard of OpenClaw. If you haven't, here's the full picture.
In late November 2025, an Austrian developer named Peter Steinberger launched an open-source AI agent originally called Clawdbot. Within weeks, it had exploded. By mid-February, it had crossed 190,000 GitHub stars and driven peak traffic of 2 million visitors in a single week, earning the title of the fastest-growing open-source project in GitHub history.
The tool works as an always-on personal assistant: It runs inside WhatsApp, Telegram and Slack.. booking flights, managing email, browsing the web for you. All on a standard Claude subscription.
|
FEATURED STORY OpenClaw: Fastest GitHub Project Ever, Now Banned by AnthropicOpenClaw was built to run on top of Claude models via subscription credentials. This week, Anthropic formally updated its terms to ban third-party tools from using those credentials. The effect: any tool piggybacking on a Claude subscription to run heavy agent workloads is now in direct violation of Anthropic's terms. The developer community pushed back hard. David Heinemeier Hansson, creator of Ruby on Rails, called the move "very customer hostile." George Hotz, founder of comma.ai, published a blog post warning Anthropic it would drive users to competing providers rather than converting them to paid API customers. Meanwhile, OpenAI CEO Sam Altman announced on X that Steinberger was joining the company to lead autonomous personal agent development. Rather than absorbing OpenClaw directly, OpenAI is sponsoring its move to an independent open-source foundation. Steinberger put it simply: "What I want is to change the world, not build a large company, and teaming up with OpenAI is the fastest way to bring this to everyone." He also added, characteristically on X, that he was joining "for the mission and because it seemed the best place to build. I'll f--- right off if that changes."
|
AI in HR? It’s happening now.
Deel's free 2026 trends report cuts through all the hype and lays out what HR teams can really expect in 2026. You’ll learn about the shifts happening now, the skill gaps you can't ignore, and resilience strategies that aren't just buzzwords. Plus you’ll get a practical toolkit that helps you implement it all without another costly and time-consuming transformation project.
The Security Picture Is Concerning
The OpenClaw excitement came with warnings that did not get enough attention.
Cybersecurity firm Kaspersky discovered nearly a thousand publicly accessible OpenClaw installations running without any authentication. SecurityScorecard found hundreds of thousands of exposed instances vulnerable to remote code execution. On ClawHub, OpenClaw's plugin marketplace, over 230 malicious plugins were uploaded in a single week. Hudson Rock disclosed a case where infostealer malware successfully extracted a victim's entire OpenClaw configuration, including API keys and chat histories.
Hackers Used Automated Code to Breach 600 Firewalls in Weeks
The second major security story this week came from Bloomberg, citing a disclosure from Amazon.
|
SECURITY ALERT 600 firewalls breached in weeks using AI-generated code Amazon disclosed this week that hackers used AI-assisted tools to automate the process of finding and exploiting firewall vulnerabilities at a scale and speed that would have previously required large, skilled teams working around the clock. The attack compromised 600 firewalls within weeks. The incident underlines a growing challenge: the same code-generation capabilities that make developers more productive can be applied directly to offensive security work, lowering the barrier for attackers considerably. |
This kind of disclosure tends to get treated as a distant, enterprise problem. But the downstream effect matters for everyone. When firewalls protecting cloud infrastructure get compromised, the data and services sitting behind them are at risk. It's not an abstract threat.
Google Puts a Music Studio Inside Gemini
This is one of the more genuinely interesting feature launches in a while.
Google announced this week that it's adding music generation to the Gemini app, powered by DeepMind's Lyria 3 model. The feature is currently in beta. You describe the song you want, and Gemini generates a 30-second track with lyrics and cover art, with the cover art made using Google's Nano Banana image model.
You can use text, images, or videos as reference material. Upload a photo of a crowded street in Tokyo, and Gemini will compose something to match the mood. You can also control style, vocals, and tempo manually.
One important design decision: you cannot directly impersonate a specific artist. If you include an artist name in your prompt, Gemini takes it as inspiration for mood or style rather than a direct copy. Google says it has filters in place to check outputs against existing content.
Every track is automatically stamped with a SynthID watermark, and Gemini now includes a tool to detect whether a piece of music was generated by its models. The feature is rolling out globally to all Gemini users aged 18 and older, with support for English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese.
Google is also expanding its Dream Track feature on YouTube, which uses Lyria 3 to help creators generate AI-composed tracks. Until this release, Dream Track was US-only. It's now available globally.
Mistral AI Buys Infrastructure Startup Koyeb
|
DEAL WATCH Mistral AI acquires Koyeb to own its deployment stack France-based Mistral AI has agreed to acquire Koyeb, a serverless software infrastructure startup, according to the Wall Street Journal. The deal gives Mistral direct control over the deployment layer its models run on - reducing dependency on third-party cloud providers and improving cost efficiency for its API customers. For a company that has positioned itself as Europe's serious alternative to OpenAI, controlling infrastructure is about more than cost savings. It's about ownership, speed, and the ability to iterate on deployment without waiting on external providers. |
World’s First Safe AI-Native Browser
AI should work for you, not the other way around. Norton Neo is the world's first safe AI-native browser with context-aware AI, built-in privacy, and configurable memory. Zero-prompt productivity that actually works.
Google and Shopee's Owner Are Building AI Shopping Together
Google and Sea Limited, the Singapore-based parent company of Shopee and gaming platform Garena, announced a new strategic partnership this week. The two companies will jointly build an AI agentic shopping prototype for Shopee, and deploy AI tools to improve game development productivity at Garena.
Shopee is not a small test bed. It held a 52% market share in Southeast Asian e-commerce in 2024, according to Momentum Works. Building agentic shopping capabilities into the platform at that scale is a meaningful deployment for Google's agent technology.
The timing is notable. Alibaba.. whose Lazada platform competes directly with Shopee across Southeast Asia, just released a new model it described as being built "for the agentic AI era." The region is becoming a genuine battleground for agentic commerce.
Reddit Is Testing AI Shopping in Search Results
Reddit announced this week that a small group of US users will start seeing a new AI-powered shopping feature inside search results. When someone searches for "best noise-canceling headphones" or "electronic gift ideas for a college student," they'll see an interactive product carousel with pricing, images, and direct buy links at the bottom of the results.
|
PLATFORM UPDATE Reddit's new AI shopping feature uses real community data What makes Reddit's approach different from a standard sponsored carousel: the products shown are pulled from real conversations. If users in a thread recommended something, that recommendation can surface as a product result. Reddit describes the goal as keeping "community perspectives at the center of the experience." Reddit CEO Steve Huffman shared some context during last week's earnings call. Weekly active users for search grew 30% over the past year, reaching 80 million. Users of Reddit Answers, the platform's AI feature, grew from 1 million in Q1 2025 to 15 million by Q4. Huffman called AI search "the next big opportunity" for the business. |
Reddit is late to the social commerce game relative to TikTok and Instagram, but its unique edge is genuine product discussion rather than curated posts. That's a real differentiator if the execution holds up.
WordPress.com Now Has a Built-In AI Assistant
Automattic launched a built-in AI assistant for WordPress.com this week. The tool sits inside the site editor and accepts natural language instructions.
You can type things like "make this section feel more modern and spacious," "change my site's colors to be brighter and bolder," or "add a contact page," and the assistant handles it directly. It also functions as a content editor: rewriting your bio, translating sections, offering headline suggestions, and checking grammar.
For image work, the assistant uses Google Gemini's Nano Banana models. A new "Generate Image" button in the Media Library lets you specify aspect ratios, styles, and content requirements to generate new images or edit existing ones.
A few practical caveats worth knowing: the tool currently only works with block themes, not classic themes. It's also opt-in, so existing users need to enable it manually under Settings > AI Tools.
India's Sarvam Launches Indus, Its First Consumer Chat App
Indian AI startup Sarvam this week launched Indus, its first consumer-facing chat app, in beta. The app is the interface for Sarvam's newly announced 105-billion-parameter model, unveiled two days earlier at the India AI Impact Summit in New Delhi.
Indus is available on iOS, Android and the web. Users can type or speak queries and receive responses in both text and audio. Sign-in works via phone number, Google, Microsoft, or Apple ID.
The competitive context gives this some weight. Sam Altman recently disclosed that India has over 100 million weekly active ChatGPT users. Anthropic reported that India accounts for 5.8% of total Claude usage, second only to the United States. Sarvam, has raised $41 million from Lightspeed, Peak XV Partners, and Khosla Ventures since its 2023 founding, is building models specifically tuned for Indian languages and use cases.
Voice-first code workflows with auto file tagging and variable recognition. Dictate reproductions and prompts and paste clean, code-friendly text into GitHub, Jira, or your editor. Try Wispr Flow for engineers.
Google AI Products: What's Live Right Now
|
GOOGLE AI SPOTLIGHT What Google has shipped or updated recently
|
The Pattern Worth Paying Attention To
Platforms are embedding AI capabilities faster than anyone is testing them. The OpenClaw situation showed what happens when a tool scales to millions of users before proper security infrastructure exists. The firewall breach showed what happens when offensive automation outpaces defensive readiness.
At the same time, there's real progress happening. Google shipping Lyria 3 as a mainstream consumer feature is not a small thing. Reddit using community intelligence to make shopping recommendations more honest matters.
It's a lot.
Stay sharp,
Better Every Day
Have a story tip or feedback? Reply to this email directly. We read everything.
