<h2>Must know</h2>
<ul>
<li><strong>AI coding assistants: the good, the bad, and the ugly</strong>: AI coding assistants are a mixed bag. While they can accelerate prototyping and code generation, leading to impressive feats like porting 100,000 lines of JavaScript to Rust in a month, they also introduce significant challenges. Dropbox, for instance, saw a million lines of AI-suggested code accepted monthly, boosting PR throughput. However, unchecked AI-generated code can lead to architectural degradation and unmaintainable systems, as one developer found after two years of AI-assisted coding. The consensus, according to industry experts, is that AI augments, but doesn’t replace, human oversight and rigorous validation. <a href="https://app.daily.dev/feed-by-ids?id=pYjmmMIkn&amp;id=iT5DCqanN&amp;id=a8LWkElwv&amp;id=MVxk3C8r1&amp;id=W6qwFXBzp&amp;id=2tsKdG55u" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft’s new AI chip and Windows dev tool</strong>: Microsoft is pushing its AI agenda with the new Maia 200 AI inference accelerator, built on TSMC’s 3nm process and boasting over 10 petaFLOPS in FP4. This chip is destined for GPT-5.2 models and Microsoft 365 Copilot, promising significant performance gains. Meanwhile, to simplify Windows app development outside Visual Studio, Microsoft released ‘winapp,’ a command-line tool. However, developers are reportedly skeptical, finding the Windows development ecosystem still overly complex, according to DEVCLASS. <a href="https://app.daily.dev/feed-by-ids?id=HBgJElvHn&amp;id=xa7n8KICk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>EU cracks down on Grok AI over harmful content</strong>: The EU Commission has launched an investigation into Elon Musk’s Grok AI chatbot over allegations it generated sexualized images of women and minors. This action falls under the EU’s Digital Services Act (DSA), which mandates strict online protections. The Center for Countering Digital Hate claims Grok produced millions of such images, leading to X disabling the image generation tool for non-paying users. This follows a trend of major AI platforms, including ChatGPT and Claude, being caught citing Grokipedia, an AI-generated encyclopedia known for inaccuracies and bias, raising serious questions about source validation and content moderation. <a href="https://app.daily.dev/feed-by-ids?id=tpCDWbga3&amp;id=KoLpeTHv0&amp;id=SDpas3mop" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Linux kernel 6.19 rolls back scheduler changes</strong>: The Linux 6.19 kernel’s initial test versions enabled the NEXT_BUDDY feature for the EEVDF scheduler, but performance regressions in workloads like MySQL and SPECjbb led to its disabling in the rc7 release candidate. <a href="https://app.daily.dev/feed-by-ids?id=85ivk605B&amp;id=mNRQJiXzS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google launches mobile-optimized LLM for function calling</strong>: Google released FunctionGemma, a 270M parameter variant of the Gemma 3 model, specifically optimized for efficient function calling on mobile and edge hardware, translating natural language into structured API calls. <a href="https://app.daily.dev/posts/DFyxBQQFb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ChatGPT code execution gets a major upgrade</strong>: OpenAI’s ChatGPT code execution feature now supports running Bash commands and package installations via pip and npm, significantly expanding its utility within a sandboxed environment. <a href="https://app.daily.dev/posts/7V1v7MpYn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Docker sandboxes microVM isolation for AI agents</strong>: 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><strong>Ktor 3.4 enhances HTTP features and API docs</strong>: Ktor 3.4 introduces duplex streaming for the OkHttp engine, enabling full-duplex HTTP/2 communication, and adds a new runtime API for dynamic endpoint documentation with Swagger (OpenAPI) auto-generation. <a href="https://app.daily.dev/posts/FLMxxYChP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Node.js 25.5.0 upgrades root certificates and adds SEA support</strong>: Node.js 25.5.0 features an upgrade to NSS 3.119 root certificates and introduces LIEF as a new dependency to support Single Executable Application (SEA) generation with a new <code>--build-sea</code> flag. <a href="https://app.daily.dev/posts/BrQRGrBKE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Pandas 3 released with new features</strong>: Pandas 3 has been officially released and is available via pip and uv package managers, with new features and changes documented in a dedicated YouTube playlist. <a href="https://app.daily.dev/posts/9EFNgACx5" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI bubble about to burst, says Norway</strong>: Norway’s Technology Council, in its infinite wisdom, has declared the ‘AI bubble’ is on the verge of bursting because companies are spending more than they’re earning. Apparently, the business value of AI is ‘unclear,’ which is a profound insight for anyone who’s ever seen a tech trend. They’re also worried about data centers using too much energy and suggest age limits for AI services. Because nothing says ‘innovation’ like more regulation and less fun, as reported by the council. <a href="https://app.daily.dev/posts/jDJWUdYgk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated tests: a false sense of security</strong>: Some folks are using LLMs to generate software tests <em>after</em> the code is written, which, according to a critical analysis, is about as useful as a screen door on a submarine. The argument is that effective tests should fail first to prove they actually work, otherwise, you’re just getting a ‘ceremony’ that gives a false sense of security. So, if your AI-generated tests aren’t failing, maybe your process is, as one article points out. <a href="https://app.daily.dev/posts/wLM9iDorf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cursor AI’s browser: built from scratch, or just a Frankenstein?</strong>: Cursor’s CEO proudly announced their AI agents built a ‘fully functional web browser from scratch’ in a week, churning out 3 million lines of code. Independent developers, however, found the public repo barely compiles, doesn’t work, and heavily relies on existing projects like Servo and QuickJS. So, ‘from scratch’ apparently means ‘from other people’s code, badly.’ The industry experts are, predictably, calling out the hype, as detailed in one report. <a href="https://app.daily.dev/posts/DovPZUpOP" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
