<h2>Must know</h2>
<ul>
<li><strong>Ingress-nginx is dead, long live Cilium</strong>: The ingress-nginx project is being archived in early 2026, meaning no more active maintenance or security updates. If you’re still using it, it’s time to migrate. You can either go for Cilium Ingress as a quick drop-in replacement or embrace Cilium’s Gateway API for more advanced features like header-based routing and cross-namespace support. Both options leverage Cilium’s eBPF-powered datapath for better performance and observability, as reported by CNCF. Migration tools are available to ease the transition. <a href="https://app.daily.dev/posts/PptBbhnAu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft Office zero-day actively exploited</strong>: Microsoft issued an emergency patch for CVE-2026-21509, a zero-day vulnerability in Office with a CVSS score of 7.8. This flaw bypasses security features for legacy COM and OLE components, allowing attackers to execute code via malicious Office files. While newer Office versions and Microsoft 365 Apps received immediate updates, Office 2016 and 2019 users are stuck with manual registry workarounds for now. CISA has mandated federal agencies patch by February 16, highlighting the urgency. <a href="https://app.daily.dev/posts/qC937DXHe" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-driven polymorphic phishing evades detection</strong>: A new phishing attack uses AI models like DeepSeek and Gemini to generate malicious JavaScript directly in victims’ browsers. The malware is fragmented and reassembled client-side, preventing static payload detection. Delivered from trusted LLM domains, this ‘last mile reassembly’ technique bypasses traditional email security and secure web gateways. User awareness is now more critical than ever, as reported by cybersecurity researchers. <a href="https://app.daily.dev/posts/jfmiKNwaJ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Google expands AI editing in Photos to new markets</strong>: Google Photos is bringing its AI-driven, prompt-based photo editing to India, Australia, and Japan, allowing users to edit images with natural language commands directly on Android devices with at least 4GB RAM, using the Nano Banana AI model. <a href="https://app.daily.dev/posts/QDh2xpUc9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google Cloud launches N4A instances with Axion ARM CPUs</strong>: Google Cloud has rolled out N4A compute instances powered by custom Axion ARM processors, offering up to 64 vCPUs and 512GB RAM, targeting web servers, microservices, and containerized applications with claims of improved performance and cost-effectiveness over previous T2A instances. <a href="https://app.daily.dev/posts/JvBjVyZ5U" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust Foundation outlines 2026-2028 strategic plan</strong>: The Rust Foundation released its 2025 Annual Report and a new strategic plan for 2026-2028, prioritizing stable infrastructure, maintainer support, responsible adoption, and community engagement, having raised $5.1 million in 2025 with $2.7 million invested back into the project. <a href="https://app.daily.dev/posts/DRpplNBlF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WhatsApp adopts Rust for media security overhaul</strong>: WhatsApp replaced 160,000 lines of C++ with 90,000 lines of Rust in its media handling stack, deploying a new media consistency library called Kaleidoscope across all platforms to validate media files and detect malicious patterns, marking the largest global deployment of Rust in a consumer app. <a href="https://app.daily.dev/posts/2YIPmtHiF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux kernel project formalizes succession plan</strong>: The Linux kernel project has formalized a succession plan for leadership after Linus Torvalds, ensuring continuity by outlining a process for maintainers to select new leadership within 72 hours of his eventual departure, addressing concerns about an aging maintainer community. <a href="https://app.daily.dev/posts/vAKZlS9Qz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GNU C Library moves to Linux Foundation infrastructure</strong>: The GNU C Library (glibc) is migrating its hosting to the Linux Foundation’s Core Toolchain Infrastructure (CTI) project to enhance cybersecurity, ensure sustainability, and diversify sponsorship, leveraging the Linux Foundation IT team’s experience with the Linux kernel. <a href="https://app.daily.dev/posts/u5a6CSvnu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MySQL faces declining community support and development instability</strong>: Despite being the second most popular database, MySQL is seeing declining community enthusiasm and development velocity due to Oracle’s layoffs and a growing feature gap between its open-source and paid versions, prompting discussions about a community fork. <a href="https://app.daily.dev/posts/u7JkpyQEs" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agents are now 'coworkers' and apparently need 'registries'</strong>: Apparently, AI agents are no longer just tools; they’re now ‘embedded coworkers’ in enterprise systems, moving from the cloud to ‘local business environments.’ This means we need ‘agent registries’ and ‘clear tiers of autonomy’ to avoid ‘agent sprawl.’ Because nothing says ‘human-centric’ like managing a bureaucracy of digital ‘coworkers’ that you can’t even fire. Who needs actual human colleagues when you can have autonomous agents causing ‘unpredictable decisions’ and ‘costly business outcomes’? <a href="https://app.daily.dev/posts/yBxtGCVqr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ChatGPT citing AI-generated encyclopedia: the circle of misinformation is complete</strong>: The Guardian found ChatGPT’s GPT-5.2 model citing Grokipedia, an AI-generated encyclopedia by Elon Musk, as a source. This recursive use of AI-generated content raises concerns about amplifying bias and misinformation, as OpenAI’s models are supposedly designed to filter out low-credibility sources. So, we’re now in a world where AI learns from AI, and the truth just becomes whatever the last AI hallucinated. What could possibly go wrong? <a href="https://app.daily.dev/posts/fWRmbZwbT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI coding agents: functional output, questionable quality</strong>: Moonshot AI’s Kimi K2.5, an open-source multimodal model, claims to outperform Gemini 3 Pro and GPT 5.2 on coding benchmarks. Yet, other reports suggest AI coding agents, like Sonnet 4.5, often produce ‘spaghetti code’ with ‘subtle semantic errors’ and ‘unnecessary complexity,’ leading to technical debt. So, while they might churn out code faster than you can say ‘refactor,’ you’ll still need a human to clean up the mess. Progress, I guess? <a href="https://app.daily.dev/feed-by-ids?id=6Szzft8Kv&amp;id=uilUGnhp4&amp;id=gcgtwbk5H" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
