<h2>Must know</h2>
<ul>
<li><strong>Microsoft patch tuesday addresses zero-days and AI vulnerabilities</strong>: Microsoft’s February 2026 Patch Tuesday is a big one, tackling 59 vulnerabilities, including six actively exploited zero-days. What’s particularly interesting for us is the fix for critical remote code execution flaws in GitHub Copilot and multiple IDEs (VS Code, Visual Studio, JetBrains), stemming from AI prompt injection vulnerabilities. These AI-related flaws pose significant risks to developers due to access to sensitive credentials and infrastructure keys, as reported by sources like Krebs on Security. <a href="https://app.daily.dev/feed-by-ids?id=zk4Rm0PBv&amp;id=BOPPPXswR&amp;id=i2YjgAdJA&amp;id=oUCth2teJ&amp;id=uxvhPzT2r&amp;id=zFYu8Wov8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents introduce new security risks and governance challenges</strong>: The rapid deployment of autonomous AI agents, expected in 40% of enterprise apps by 2026, brings significant security vulnerabilities and policy concerns. GitGuardian, which just raised $50M, is focusing on securing these ‘Non-Human Identities’ and addressing gaps like prompt injection. Meanwhile, researchers are already demonstrating command hijacking attacks against embodied AI, highlighting the urgent need for robust security frameworks as these agents gain more autonomy. <a href="https://app.daily.dev/feed-by-ids?id=0rQwEgnbB&amp;id=QRBfTNO1G&amp;id=ZVqULJLz3&amp;id=gUwIhEnZO&amp;id=l8YCDT7qf&amp;id=mmQEWnSC4&amp;id=BOPPPXswR&amp;id=i2YjgAdJA&amp;id=oUCth2teJ&amp;id=uxvhPzT2r" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Open-source GLM-5 model claims parity with proprietary AI</strong>: <a href="http://Z.ai" target="_blank" rel="noopener nofollow">Z.ai</a> has released GLM-5, a new open-source large language model under an MIT License, claiming it matches proprietary models like Claude Opus 4.6 and GPT 5.3 Codex for long-horizon coding tasks. With 744B parameters and a novel asynchronous RL infrastructure, it’s available for free access via Modal until April, integrating with popular AI frontends and coding agents. This could be a significant development for ‘vibe engineering’ and multi-agent AI systems. <a href="https://app.daily.dev/posts/bOspGaOnW" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>WebMCP standard proposed for AI agent interaction</strong>: Chrome is previewing WebMCP, a new web standard that allows websites to expose structured tools directly to in-browser AI agents, moving beyond brittle DOM scraping. <a href="https://app.daily.dev/feed-by-ids?id=Ib422mkez&amp;id=NpNZ22j1b&amp;id=R0k8YNGTx" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered testing tools emerge</strong>: Datadog launched Code Coverage with AI-generated tests, and GitHub Copilot Testing for .NET is now generally available, indicating a strong trend towards AI-driven automation in software testing. <a href="https://app.daily.dev/feed-by-ids?id=gZJqHPom7&amp;id=j254S16VC&amp;id=sJ5P2kSW8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks launches lakeflow for data engineering</strong>: Databricks introduced Lakeflow on Azure, a unified data engineering solution for ingestion, transformation, orchestration, and governance, featuring AI-powered debugging and serverless compute. <a href="https://app.daily.dev/feed-by-ids?id=Bf1dK14FP&amp;id=DckO36Vaq&amp;id=cW6KgKgEK&amp;id=omSTWjEfj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Erlang/OTP 29.0-rc1 released</strong>: Erlang/OTP 29.0-rc1 introduces native records, multi-valued comprehensions, and quantum-resistant SSH key exchange, while dropping 32-bit Windows builds. <a href="https://app.daily.dev/posts/fTCktIrzF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>FastAPI cloud simplifies deployment</strong>: FastAPI Cloud, backed by Sequoia, offers single-command deployment for FastAPI applications, handling HTTPS, auto-scaling, and infrastructure complexity on Kubernetes. <a href="https://app.daily.dev/feed-by-ids?id=5Mzv43k5F&amp;id=5atbRAfWV&amp;id=6Rrig1lfL&amp;id=7B5EDflI8&amp;id=N2p8QaBAl" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Pandas 3.0.0 introduces breaking changes</strong>: Pandas 3.0.0 is out with a dedicated string dtype, formal Copy-on-Write semantics, and early support for declarative column transformations, requiring Python 3.11+. <a href="https://app.daily.dev/feed-by-ids?id=Bf1dK14FP&amp;id=DckO36Vaq&amp;id=cW6KgKgEK&amp;id=omSTWjEfj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI’s ChatGPT atlas browser architecture detailed</strong>: OpenAI’s ChatGPT Atlas browser uses a unique architecture that runs Chromium as a separate process for instant startup and responsiveness with hundreds of tabs, supporting rich SwiftUI animations. <a href="https://app.daily.dev/posts/TsFQIm4Vj" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI will make you obsolete, unless you buy our tools</strong>: One ‘industry leader’ is screaming that 50% of white-collar jobs, including engineering, will be gone in 1-5 years thanks to AI, urging everyone to immediately adopt paid AI tools and experiment daily. Because, apparently, the only way to survive the AI apocalypse is to buy their stuff and frantically ‘vibe code’ your way to relevance. Sounds like a classic fear-mongering sales pitch to me. <a href="https://app.daily.dev/feed-by-ids?id=jQIyW7C3o&amp;id=lGyFb7XTB&amp;id=waJpUtc5V" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Replit’s ‘build an app in minutes with AI’ demo</strong>: Replit is showcasing a mobile app builder that lets you create and publish apps with natural language prompts, claiming it takes ‘minutes’ without traditional coding. While impressive for a demo, the real world of app development, debugging, and deployment is rarely a ‘minutes’ affair, especially when you’re trying to avoid ‘vibe coding’ breaking your entire system. <a href="https://app.daily.dev/posts/qWp2u9GdQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The death of traditional testing, again</strong>: Apparently, ‘agentic development has broken traditional testing paradigms,’ and LLM-generated JiTTests are here to ‘auto-heal’ your tests and eliminate maintenance costs. Because, you know, humans are just too slow and error-prone for something as critical as ensuring code quality. Let’s just trust the AI to test the AI-generated code, what could possibly go wrong? <a href="https://app.daily.dev/feed-by-ids?id=gZJqHPom7&amp;id=j254S16VC&amp;id=sJ5P2kSW8" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
