<h2>Must know</h2>
<ul>
<li>AI coding assistants increase planning demands and security risks: AI coding assistants, while boosting code writing speed, demand significantly more upfront architectural planning and specification work. A 2025 study found developers using AI assistants took 19% longer to complete tasks overall, despite feeling faster. Furthermore, 45% of AI-generated code contains security vulnerabilities, and code review times on high-AI-usage teams have increased by 91%. This suggests that treating AI as a junior developer needing clear direction, rather than an autonomous agent, is crucial to avoid an ‘architecture tax’ and maintain code quality, as reported by multiple sources. <a href="https://app.daily.dev/posts/pYjmmMIkn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>TSMC CapEx strategy creates AI chip supply bottleneck for hyperscalers: TSMC’s flat capital expenditures from 2021 to 2024, despite a surge in hyperscaler AI investments, have led to a significant AI chip supply bottleneck. This imbalance is projected to cost hyperscalers billions in lost AI revenue and is expected to persist through at least 2028. The analysis suggests hyperscalers should prioritize supply diversification, even if it means shifting away from TSMC, to mitigate ongoing revenue risks. <a href="https://app.daily.dev/posts/pPxEIulIE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Linux kernel 6.19 rolls back EEVDF scheduler feature due to performance regressions: The Linux 6.19 kernel’s initial test versions enabled the NEXT_BUDDY feature for the EEVDF scheduler, but quickly faced reports of performance regressions in workloads like MySQL and SPECjbb. These issues led to the feature being disabled in the rc7 release candidate, highlighting the kernel community’s rapid response to performance impacts. Benchmarks confirmed the regressions were directly tied to NEXT_BUDDY, leading to its temporary removal. <a href="https://app.daily.dev/feed-by-ids?id=85ivk605B&amp;id=mNRQJiXzS" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li>Docker Sandboxes introduces microVM isolation for AI coding agents: Docker Sandboxes now offers microVM-based isolation for AI coding agents like Claude Code and Gemini CLI, allowing them to operate securely without impacting the host system, currently available on macOS with Windows and Linux support planned. <a href="https://app.daily.dev/posts/6YdOupFK7" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>OpenAI upgrades ChatGPT code execution with Bash and package installation: OpenAI’s ChatGPT code execution feature now supports running Bash commands and installing packages via pip and npm within its sandboxed environment, significantly expanding its utility for all users, including free-tier. <a href="https://app.daily.dev/posts/7V1v7MpYn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Microsoft launches Maia 200 AI inference accelerator: Microsoft has unveiled the Maia 200, a custom AI inference accelerator built on TSMC’s 3nm process, featuring FP8/FP4 tensor cores and 216GB HBM3e memory, claiming 30% better performance per dollar for tasks like GPT-5.2 models and Microsoft 365 Copilot. <a href="https://app.daily.dev/posts/HBgJElvHn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AMDGPU Linux driver adds HDMI VRR and ALLM gaming support: New AMDGPU Linux kernel driver patches introduce HDMI Variable Rate Refresh (VRR) and Auto Low Latency Mode (ALLM) for gaming, aiming for feature parity with DisplayPort and Windows, despite obstacles from the HDMI Forum’s closed standards. <a href="https://app.daily.dev/posts/bsF1QAXcQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Node.js 25.5.0 released with security and feature enhancements: Node.js 25.5.0 is out, featuring an upgrade to NSS 3.119 root certificates, a new <code>--build-sea</code> flag for Single Executable Application generation, and an <code>ignore</code> option for <code>fs.watch</code>, alongside SQLite defensive mode and <code>test_runner</code> improvements. <a href="https://app.daily.dev/posts/BrQRGrBKE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>KDE Plasma 6.6 shifts to systemd, ends FreeBSD support: KDE Plasma 6.6 beta introduces a systemd-logind dependent login manager, effectively ending FreeBSD support and signaling an increasing Linux-specific direction for the desktop environment, mirroring GNOME’s similar shift. <a href="https://app.daily.dev/posts/satawafBY" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Uncloud Kubernetes alternative gains momentum among small teams: Uncloud, an open-source deployment platform leveraging Docker, Caddy, and WireGuard, is gaining traction as a simpler Kubernetes alternative for small teams, offering single-command deployments and nearing a stability milestone with plans for a managed control center. <a href="https://app.daily.dev/posts/FrS1fa85P" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li>Cursor AI agents’ browser claims debunked: Cursor’s CEO proudly announced their AI agents built a ‘fully functional web browser from scratch’ in a week, generating over 3 million lines of code. Turns out, independent developers found the browser barely compiles, relies heavily on existing projects like Servo and QuickJS, and performs ‘abysmally.’ So much for ‘from scratch’ and ‘functional’—just another AI hype cycle, as detailed by critical analysis of the public GitHub repository. <a href="https://app.daily.dev/posts/DovPZUpOP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI-generated software tests: a false sense of security: Apparently, using LLMs to generate software tests after the code is written is all the rage. But according to some, this just creates a ‘ceremony’ that gives a false sense of security. The real trick, they say, is to make sure your tests can actually fail to prove they’re useful. Otherwise, you’re just admiring AI’s ability to write more code, not better code. <a href="https://app.daily.dev/posts/wLM9iDorf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI coding assistants: the silent killer of maintainable code: One developer, after two years of AI-assisted coding, discovered that while AI-generated code looked good on paper and passed reviews, it led to ‘cumulative architectural degradation’ and an unmaintainable codebase. The AI, it seems, is great at individual tasks but terrible at understanding the bigger picture. So, if your code is becoming ‘slop,’ maybe it’s time to ditch the AI and go back to doing it yourself, because apparently, that’s faster in the long run. <a href="https://app.daily.dev/posts/MVxk3C8r1" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>