<h2>Must know</h2>
<ul>
<li>AI security and privacy concerns escalate: The rapid deployment of AI is raising serious security and privacy red flags. Researchers found over 175,000 Ollama AI hosts exposed online without safety guardrails, creating potential for resource hijacking and remote attacks. SpaceX also updated Starlink’s privacy policy to use customer data for AI training, coinciding with talks to merge with xAI, sparking concerns about data misuse. Additionally, a critical 1-click RCE vulnerability was discovered in OpenClaw, an AI assistant used by over 100,000 developers, highlighting the inherent risks in rapidly adopted AI tools. <a href="https://app.daily.dev/feed-by-ids?id=2SdHHyeRJ&amp;id=YUSdAM3D6&amp;id=d76FYhLFT&amp;id=iEIaVGrhd&amp;id=yXHHUa4yo" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI code generation: faster but riskier: AI code generation can accelerate production by 55% but shifts bottlenecks to verification, increasing review times by 91% in high-AI-usage teams. Microsoft’s increased emergency patches and claims of 30% AI-contributed code suggest a correlation with potential quality issues. AI-generated code often lacks system context, generates non-validating tests, and erodes architectural resilience, leading to increased code churn and minimal gains in legacy codebases. This points to a need for robust guardrails and platform engineering to manage the influx of AI-generated code effectively. <a href="https://app.daily.dev/feed-by-ids?id=CUMsxmpoK&amp;id=H1aWwdchu&amp;id=KYToiOjQs&amp;id=TCb4dVj23" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Cloud native ecosystem sees major shifts: The cloud native ecosystem is undergoing significant changes, with the Ingress NGINX project announcing its retirement and the Kubernetes Dashboard project being archived. This signals a move towards new solutions and potentially more specialized tools. On a positive note, Cluster API v1.12 was released with in-place updates, and a new Checkpoint/Restore Working Group has been formed, indicating continued evolution and refinement in core cloud infrastructure, as reported by CloudNative.Now. <a href="https://app.daily.dev/posts/EdFbm39PB" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li>New open-source Kubernetes dashboard radar released: A new open-source Kubernetes dashboard called Radar, built as a single Go binary with an embedded React frontend, provides real-time cluster visibility without cloud logins or agents, supporting GitOps tools like Argo CD and Flux. <a href="https://app.daily.dev/feed-by-ids?id=DiEunA7dc&amp;id=a3wJg58hN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>NVIDIA NVFP4 quantization for LLMs released by Red Hat: Red Hat has released NVFP4-quantized versions of popular LLMs, leveraging NVIDIA’s 4-bit floating-point format for efficient deployment with near-baseline accuracy, especially effective for frontier-scale models. <a href="https://app.daily.dev/posts/OPe22PWoI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>MLflow integrates new evaluation metrics for LLMs and RAG: MLflow 3.9.0 now integrates over 50 evaluation metrics from DeepEval, RAGAS, and Arize Phoenix, providing a unified API to assess LLM agents and RAG systems for conversational quality, retrieval, hallucination, and safety. <a href="https://app.daily.dev/feed-by-ids?id=3L44pW4ot&amp;id=5ZJWT4v6q&amp;id=JVTlCrLRm&amp;id=h7CGoaIP0&amp;id=zVcgznCs0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Wix introduces xEngineer model for AI-driven development: Wix is integrating AI into engineering roles with its ‘xEngineer’ model, aiming for engineers to take broader ownership across the development lifecycle, supported by AI tools to reduce cognitive load and focus on judgment. <a href="https://app.daily.dev/posts/1ESLU5xBs" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Elixir forum celebrates 10 years and plans for LLM era: The Elixir Forum marked its 10th anniversary with significant growth and announced structural changes to adapt to Large Language Models and evolving user behavior, including a shift from its ‘single-thread-per-library’ rule. <a href="https://app.daily.dev/posts/lMXgCoFLh" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>WebAssembly celebrates 10 years, expands beyond browser: WebAssembly (WASM) marked its 10th anniversary, evolving from a browser performance solution to a foundational technology for edge platforms, serverless runtimes, and embedded systems, with growing promise for AI workloads. <a href="https://app.daily.dev/posts/XjOBMuNCm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>DataDog releases pg_tracing for PostgreSQL distributed tracing: DataDog introduced <code>pg_tracing</code>, a PostgreSQL extension for server-side distributed tracing that generates spans for database queries and operations, supporting versions 14-16 and OpenTelemetry integration. <a href="https://app.daily.dev/posts/Xfr0u4GSj" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li>Native PHP makes mobile development ‘completely free and open source’: Native PHP just dropped some ‘major updates’ making mobile development ‘completely free and open source,’ which sounds fantastic until you realize it’s a single, uncorroborated announcement. They’re touting AI tools like Mimi to ‘rapidly generate complete Laravel mobile applications’ from natural language. Because, you know, building complex mobile apps is just a matter of typing a few words and poof, production-ready code appears, completely free of charge and maintenance. <a href="https://app.daily.dev/posts/bRf9BXT9O" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI will make tiny teams outcompete giants, apparently: According to some ‘experts,’ a ‘significant divide’ is emerging between AI users, where ‘power users’ with tools like Claude Code will enable ‘tiny teams’ to ‘outcompete organizations thousands of times their size.’ Meanwhile, enterprises are stuck with ‘poor tools like M365 Copilot’ and ‘locked-down IT policies.’ So, the future of tech dominance hinges on who has the most flexible IT and the best AI chat client, not, you know, actual product-market fit or business acumen. <a href="https://app.daily.dev/posts/WZuX1OWTx" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI-generated historical videos are a hit, said no one ever: Darren Aronofsky and TIME magazine, with Google DeepMind’s tech, released an AI-generated video series called ‘On This Day… 1776,’ and it was met with ‘widespread criticism and negative reception.’ This ‘continues a pattern of AI-generated visuals failing to resonate with audiences.’ But hey, at least they tried to commemorate the American Revolution with blurry, uncanny valley visuals, proving once again that just because you can generate something with AI, doesn’t mean you should. <a href="https://app.daily.dev/feed-by-ids?id=0hvZGEXEr&amp;id=vgyq27hp2" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>