<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. Critically, it also patches 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 Krebs on Security. <a href="https://app.daily.dev/posts/zk4Rm0PBv" 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 governance concerns. Cisco introduced AI BOM, an open-source tool for inventorying AI assets, while GitGuardian secured $50M to address non-human identity (NHI) crisis and AI agent security gaps. Experts warn that competitive pressures are leading to rapid deployment without adequate safety frameworks, as highlighted by GitGuardian. <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=UhlBeqTGk&amp;id=BOPPPXswR&amp;id=i2YjgAdJA&amp;id=oUCth2teJ" 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> released GLM-5, a new open-source large language model under an MIT License, claiming to match proprietary models like Claude Opus 4.6 and GPT 5.3 Codex for long-horizon coding tasks. With 744B parameters and novel asynchronous RL infrastructure, it aims for best-in-class performance among open-source models on reasoning, coding, and agentic tasks. Modal is offering free access via an OpenAI-compatible API endpoint until the end of April. <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>AI-powered testing tools emerge for code quality</strong>: Datadog launched Code Coverage with AI-generated tests, while GitHub Copilot Testing for .NET is now generally available, offering AI-powered unit test generation and coverage summaries. <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>Google Cloud launches GEAR program for enterprise AI agents</strong>: Google Cloud introduced the GEAR (Gemini Enterprise Agent Ready) program, a structured learning pathway for building and deploying AI agents using their Agent Development Kit. <a href="https://app.daily.dev/feed-by-ids?id=9C4SjPZli&amp;id=gwxHevbDX&amp;id=o13HbD6JR&amp;id=r88T1MoRe&amp;id=vS730ONZY" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks launches Lakeflow for unified data engineering</strong>: Databricks announced Lakeflow on Azure Databricks, a unified solution for data 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>OpenAI details ChatGPT Atlas browser architecture</strong>: OpenAI’s ChatGPT Atlas browser uses a unique architecture that runs Chromium as a separate process, enabling instant startup, responsiveness with hundreds of tabs, and rich SwiftUI animations. <a href="https://app.daily.dev/posts/TsFQIm4Vj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vercel Flags now in public beta</strong>: Vercel Flags is now publicly available, offering native feature flag management directly in the Vercel Dashboard with SDK support for Next.js and SvelteKit, and OpenFeature standard integration. <a href="https://app.daily.dev/posts/IfJOyXyyM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>FastAPI Cloud simplifies application 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>Erlang/OTP 29.0-rc1 released with new features</strong>: Erlang/OTP 29.0-rc1 introduces native records, multi-valued comprehensions, and quantum-resistant SSH key exchange, while removing 32-bit Windows builds. <a href="https://app.daily.dev/posts/fTCktIrzF" 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 my tools</strong>: One ‘industry leader’ is screaming that 50% of white-collar jobs will vanish 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 click buttons, lest you miss the ‘closing window’ of competitive advantage. 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>AI agents building C compilers for $20k</strong>: Anthropic apparently spent $20,000 and unleashed 16 Claude AI agents to build a C compiler. The result? A 100,000-line codebase that struggles with basic compilation, relies on GCC for critical parts, and generates less efficient code. So, for the price of a decent used car, you too can have an AI-generated compiler that’s worse than what a human could write in a weekend. Progress! <a href="https://app.daily.dev/posts/D7CXYuXPP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI will fix all your data problems, just trust us</strong>: Arango is here to tell us that vector databases are ‘insufficient’ for enterprise AI and that only their ‘context-first data layer’ can unify meaning, time, and trust. Because, you know, all those fragmented architectures are just too much work. They’re basically saying, ‘Your data is a mess, but don’t worry, our magic AI solution will make it all better.’ Sure, Jan. <a href="https://app.daily.dev/feed-by-ids?id=0lLQRkBLb&amp;id=Zm3R4GzeA&amp;id=bYavSTao6&amp;id=daHpBJ83O&amp;id=j9QS83E6P&amp;id=kLzDfQ5ZL" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
