<h2>Must know</h2>
<ul>
<li><strong>AI code generation: faster code, slower everything else</strong>: AI code generation can speed up coding by 55%, but it shifts bottlenecks to verification, increasing review times by 91% in high-AI-usage teams. Microsoft, for example, is seeing more emergency patches and claims 30% of its code is AI-generated, suggesting a correlation. The real issue is AI optimizing code without system context, leading to non-validating tests, documenting non-existent features, and eroding architectural resilience, as detailed in multiple reports. <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><strong>Cloud native ecosystem sees major shifts</strong>: The cloud native ecosystem is undergoing significant changes, with the Ingress NGINX project announcing its retirement and the Kubernetes Dashboard project being archived. On a brighter note, Cluster API v1.12 was released with in-place updates, and a new open-source Kubernetes dashboard called Radar, built with Go and React, aims to provide real-time cluster visibility without agents or cloud logins, addressing limitations of existing tools like Lens and k9s. <a href="https://app.daily.dev/feed-by-ids?id=EdFbm39PB&amp;id=DiEunA7dc&amp;id=a3wJg58hN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI security and privacy concerns escalate</strong>: AI security is becoming a minefield. Researchers found over 175,000 Ollama AI hosts exposed online, running models without safety guardrails, creating potential for resource hijacking and remote attacks. SpaceX updated Starlink’s privacy policy to use customer data for AI training, coinciding with talks to merge with xAI. Meanwhile, a critical 1-click RCE vulnerability was found 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&amp;id=4nlzVZMEL&amp;id=ALGXJz29U&amp;id=fuL6WfKHt&amp;id=oAoGgXAAn&amp;id=zgJdriYuf" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>NVIDIA NVFP4 quantization released by Red Hat</strong>: Red Hat 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><strong>MLflow integrates new LLM evaluation metrics</strong>: 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 with specialized metrics for conversational agents and hallucination detection. <a href="https://app.daily.dev/posts/zVcgznCs0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Wix introduces xEngineer model for AI-driven development</strong>: Wix launched its ‘xEngineer’ model, integrating AI into engineering roles to broaden ownership across the development lifecycle, aiming to reduce cognitive load and focus engineers on judgment rather than mechanical analysis. <a href="https://app.daily.dev/posts/1ESLU5xBs" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WebAssembly celebrates 10 years, expands beyond browser</strong>: WebAssembly (WASM) marked its 10th anniversary, evolving into a foundational technology for edge platforms, serverless runtimes, and embedded systems, with growing promise for AI workloads at the edge. <a href="https://app.daily.dev/posts/XjOBMuNCm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New Svelte type checker claims 24x speedup</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 a Go-based TypeScript compiler. <a href="https://app.daily.dev/posts/6aMfl6kdf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>DataDog releases pg_tracing for PostgreSQL</strong>: DataDog developed <code>pg_tracing</code>, a PostgreSQL extension for server-side distributed tracing that generates spans for database queries and operations, supporting PostgreSQL 14-16 and OpenTelemetry. <a href="https://app.daily.dev/posts/Xfr0u4GSj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux kernel 6.19 to include Uniwill/TUXEDO driver</strong>: The upcoming Linux 6.19 kernel will integrate a new Uniwill laptop platform driver, enhancing hardware features and settings, including cTGP graphics power controls, for Uniwill-manufactured and TUXEDO laptops. <a href="https://app.daily.dev/posts/mzBRRS79C" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-washing: the new corporate scapegoat</strong>: Companies are blaming AI for over 50,000 layoffs in 2025, but a Forrester report suggests this is mostly ‘AI-washing’—using AI as an excuse for cuts driven by financial pressures or over-hiring, according to TechCrunch. Turns out, most of these companies don’t even have mature AI applications ready to replace those roles. Who knew AI could be such a convenient scapegoat for bad business decisions? <a href="https://app.daily.dev/posts/yXHHUa4yo" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI’s ChatGPT: web search optional, accuracy not so much</strong>: OpenAI makes it surprisingly difficult to enable web search in ChatGPT, despite the model frequently hallucinating without it. The UX actively discourages search, leading to questions about whether OpenAI prioritizes cost savings over accuracy. So, while they’re busy touting their advanced models, users are left playing hide-and-seek with basic factual accuracy. <a href="https://app.daily.dev/posts/0hvZGEXEr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated video: still not good</strong>: Darren Aronofsky and TIME magazine released an AI-generated video series, ‘On This Day… 1776,’ using Google DeepMind tech, and it was met with widespread criticism and negative reception. This just confirms the pattern: AI-generated visuals continue to fail at resonating with mainstream audiences. Maybe some things still need a human touch, or at least a human eye for what doesn’t look like a fever dream. <a href="https://app.daily.dev/posts/vgyq27hp2" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
