<h2>Must know</h2>
<ul>
<li><strong>WhatsApp swaps C++ for Rust for media security</strong>: WhatsApp has replaced 160,000 lines of C++ code with 90,000 lines of Rust in its media handling stack, deploying a new library called Kaleidoscope. This change, affecting billions of users across all platforms, aims to validate media files against format standards and detect malicious patterns to prevent malware exploits, as reported by Facebook Engineering. Meta claims improved performance and memory usage, showcasing Rust’s production-readiness for large-scale applications. <a href="https://app.daily.dev/posts/2YIPmtHiF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft Office zero-day vulnerability 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, actively exploited in the wild. The flaw bypasses security features for legacy COM and OLE components, allowing attackers to execute code via malicious Office files. While newer Office versions received immediate updates, Office 2016 and 2019 users need manual registry workarounds, with patches still pending. CISA has mandated federal agencies patch by February 16, highlighting the urgency. <a href="https://app.daily.dev/feed-by-ids?id=qC937DXHe&amp;id=qC937DXHe" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google expands Android anti-theft features</strong>: Google is rolling out new anti-theft features for Android 16 and above, including a Failed Authentication Lock, expanded Identity Check coverage for biometric-protected apps, and an optional security challenge for Remote Lock. In Brazil, Theft Detection Lock, an AI-powered feature detecting snatch-and-run motions, will be enabled by default, along with Remote Lock. These updates aim to strengthen security and theft prevention, though details on effectiveness and broader rollout are not specified. <a href="https://app.daily.dev/posts/a8ZBOybpX" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Linux kernel formalizes succession plan</strong>: The Linux kernel project has formalized a succession plan for leadership after Linus Torvalds’ eventual departure, documented in official kernel documentation to ensure project sustainability and operational continuity. <a href="https://app.daily.dev/posts/vAKZlS9Qz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ingress-nginx project to be archived, migration to Cilium recommended</strong>: The ingress-nginx project will be archived in early 2026, ending active maintenance and security updates, with users advised to migrate to Cilium Ingress or Cilium’s Gateway API for continued support and enhanced features. <a href="https://app.daily.dev/posts/PptBbhnAu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust Foundation releases 2025 annual report and strategic plan</strong>: The Rust Foundation released its 2025 Annual Report and a 2026-2028 strategic plan, prioritizing stable infrastructure, maintainer support, responsible adoption growth, and community engagement, having raised $5.1 million in 2025. <a href="https://app.daily.dev/posts/DRpplNBlF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google adds visual reasoning to Gemini 3 Flash with Agentic Vision</strong>: Google has introduced Agentic Vision capabilities to its Gemini 3 Flash model, allowing it to interactively analyze images, merge visual reasoning with code execution, and generate visualizations from complex data, accessible via Google AI Studio and Vertex AI. <a href="https://app.daily.dev/posts/KIgS0LnaZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-driven polymorphic phishing uses LLMs for dynamic code generation</strong>: A new phishing attack uses AI models like DeepSeek and Gemini to generate malicious JavaScript client-side, splitting malware into fragments delivered from trusted LLM domains to evade traditional detection systems. <a href="https://app.daily.dev/posts/jfmiKNwaJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks SQL sees significant performance and cost improvements</strong>: Databricks SQL achieved up to 40% performance improvement in 2025 through automatic engine optimizations, alongside 10x reduced Unity Catalog latency, 30% faster Delta Sharing queries, and 40% storage cost reduction via Zstandard compression. <a href="https://app.daily.dev/feed-by-ids?id=rT9ZMYOFd&amp;id=rT9ZMYOFd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS Marketplace now supports self-service FPGA AMI listings</strong>: AWS Marketplace now allows vendors to self-list AMI products containing FPGA images, streamlining the process for hardware accelerator vendors by eliminating manual forms and supporting up to 15 Amazon FPGA images per listing. <a href="https://app.daily.dev/feed-by-ids?id=HO7u7zIEu&amp;id=HO7u7zIEu" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agents built a browser from scratch, or so they say</strong>: Cursor announced its AI agents built a web browser with millions of lines of code and a custom Rust rendering engine, touting it as a major AI breakthrough. However, developers who actually looked at the code found it to be ‘spaghetti,’ unable to compile, and heavily reliant on existing external projects, completely undermining the ‘from scratch’ claim. Apparently, ‘AI-generated’ now means ‘we glued together a bunch of open-source stuff and called it new.’ <a href="https://app.daily.dev/feed-by-ids?id=gcgtwbk5H&amp;id=gcgtwbk5H&amp;id=gcgtwbk5H" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cloudflare’s ‘serverless’ Matrix homeserver is ‘fundamentally broken’</strong>: Cloudflare proudly announced a ‘serverless’ Matrix homeserver on its Workers platform, claiming zero operational overhead and post-quantum security. But Matrix developers quickly pointed out that it lacks basic security features like authorization checks and state resolution, calling the implementation ‘fundamentally broken’ and possibly ‘AI-generated.’ So much for ‘secure’ and ‘interoperable’ when it can’t even handle message history or talk to other Matrix servers. <a href="https://app.daily.dev/feed-by-ids?id=93HDTbjA4&amp;id=93HDTbjA4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ChatGPT caught citing Elon Musk’s AI-generated encyclopedia</strong>: ChatGPT’s GPT-5.2 model was caught citing Grokipedia, an AI-generated encyclopedia by Elon Musk, as a source in its responses. The Guardian found multiple instances, raising concerns about AI systems recursively referencing each other, potentially amplifying bias and misinformation. OpenAI claims its models filter low-credibility sources, but apparently, ‘AI-generated by a billionaire’ passes the sniff test. <a href="https://app.daily.dev/posts/fWRmbZwbT" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
