<h2>Must know</h2>
<ul>
<li><strong>Kubernetes RCE flaw found in common permissions</strong>: A critical design flaw in Kubernetes allows remote code execution (RCE) in any container on a node via commonly used <code>nodes/proxy GET</code> permissions. This issue, while not a CVE, poses a significant risk, especially since many monitoring tools and network overlays utilize these permissions. A fix is slated for Kubernetes 1.36, but given the slow update cycles in many production environments, organizations are urged to implement fine-grained authorization controls and scan for RCE attempts. <a href="https://app.daily.dev/posts/EdFbm39PB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI security risks escalate with exposed models and data privacy changes</strong>: Researchers found over 175,000 Ollama AI hosts exposed online without safety guardrails, enabling potential resource hijacking and remote execution attacks. Concurrently, SpaceX updated Starlink’s privacy policy to use customer data for AI training, coinciding with talks to merge with xAI. This raises significant concerns about data misuse and the security of AI deployments, as reported by sources like TechCrunch and CSO Online. <a href="https://app.daily.dev/feed-by-ids?id=2SdHHyeRJ&amp;id=YUSdAM3D6&amp;id=d76FYhLFT&amp;id=iEIaVGrhd&amp;id=yXHHUa4yo&amp;id=ALGXJz29U&amp;id=O6F9FNK2K&amp;id=dF5gFHPqw&amp;id=gnIYjeUsK&amp;id=jjjtJaG53" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Notepad++ compromised by state-sponsored hackers</strong>: Notepad++ was compromised by suspected state-sponsored hackers for six months, with attackers intercepting update traffic and redirecting users to malicious servers. The breach, which exploited insufficient update verification, highlights the vulnerability of widely used open-source tools. Remediation involved migrating to a new hosting provider and implementing enhanced security measures in recent Notepad++ versions. <a href="https://app.daily.dev/posts/NmnSVgxR6" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Wix integrates AI into engineering roles with xEngineer model</strong>: Wix introduced the “xEngineer” model, integrating AI into engineering roles to broaden ownership across the development lifecycle, particularly in DevOps and SRE, aiming to evolve responsibilities rather than reduce headcount. <a href="https://app.daily.dev/posts/1ESLU5xBs" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New open-source Kubernetes dashboard Radar released</strong>: A new open-source Kubernetes dashboard called Radar was introduced, built as a single Go binary with an embedded React frontend, providing real-time cluster visibility without cloud logins or cluster-side agents. <a href="https://app.daily.dev/feed-by-ids?id=DiEunA7dc&amp;id=a3wJg58hN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA NVFP4 quantization released for LLMs</strong>: NVIDIA released NVFP4, a 4-bit floating-point quantization format, with Red Hat providing NVFP4-quantized versions of popular LLMs, significantly reducing storage while maintaining high accuracy on NVIDIA Blackwell GPUs. <a href="https://app.daily.dev/posts/OPe22PWoI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MLflow integrates new evaluation metrics for LLM agents</strong>: MLflow 3.9.0 now integrates over 50 evaluation metrics from DeepEval, RAGAS, and Arize Phoenix frameworks, offering a unified API to evaluate LLM agents and RAG systems with specialized metrics for conversational agents and hallucination detection. <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><strong>Linux kernel 6.19 to add Uniwill laptop driver</strong>: The Linux 6.19 kernel will integrate a new Uniwill laptop platform driver, enabling cTGP graphics power settings and other platform-specific controls for Uniwill-manufactured and TUXEDO Computers models. <a href="https://app.daily.dev/posts/mzBRRS79C" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WebAssembly celebrates 10 years, expands to edge and AI</strong>: WebAssembly (WASM) celebrated its 10th anniversary, evolving beyond browser performance to become a foundational technology for edge platforms, serverless runtimes, and embedded systems, promising to enable composable AI agent systems. <a href="https://app.daily.dev/posts/XjOBMuNCm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Svelte gets new fast type checker</strong>: An experimental type and compiler warning checker, <code>svelte-fast-check</code>, was announced for Svelte and SvelteKit projects, claiming up to 24x faster performance than <code>svelte-check</code> by leveraging TypeScript 7 (tsgo). <a href="https://app.daily.dev/posts/6aMfl6kdf" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI blamed for layoffs, or is it just ‘AI-washing’?</strong>: Apparently, AI is now the scapegoat for corporate layoffs, with over 50,000 jobs cut in 2025 blamed on our silicon overlords. However, a Forrester report suggests this is mostly ‘AI-washing,’ a convenient excuse for financial pressures or pandemic over-hiring, because, you know, it sounds better than ‘we messed up.’ <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><strong>OpenAI discourages web search in ChatGPT, prioritizes hallucination</strong>: OpenAI seems to be actively discouraging web search in ChatGPT, making it difficult to enable despite the model’s penchant for making things up. It’s almost as if they’re prioritizing cost savings over, you know, accuracy. Who needs facts when you have a chatbot that confidently hallucinates? <a href="https://app.daily.dev/feed-by-ids?id=0hvZGEXEr&amp;id=vgyq27hp2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Norwegian developer predicts AI bubble burst</strong>: A Norwegian developer, clearly a prophet of doom, predicts the AI bubble will burst in 2026, leading to higher prices and lower quality AI services. His advice? Avoid proprietary AI models and maintain core programming skills. Because, apparently, AI-generated code lacks ‘domain understanding’ and creates ‘unmaintainable systems.’ Shocking, I know. <a href="https://app.daily.dev/posts/NEIMwAdHs" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
