<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 detailed specifications. A 2025 study revealed that teams using AI assistants took 19% longer to complete tasks overall, despite feeling faster. This is due to issues like AI lacking persistent memory, leading to ‘context pollution’ and increased debugging time, and a staggering 45% of AI-generated code containing security vulnerabilities, as reported by an industry analysis. Proper use requires treating AI as a junior developer needing clear direction, not an autonomous agent. <a href="https://app.daily.dev/posts/pYjmmMIkn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>OpenAI upgrades ChatGPT code execution with Bash and package installs: OpenAI has significantly enhanced ChatGPT’s code execution feature, now allowing users to run Bash commands and install packages via pip and npm within its sandboxed environment. This upgrade, which also includes a container.download tool for web file retrieval, expands potential use cases across multiple programming languages like JavaScript, Ruby, and C++. The new capabilities are available to all ChatGPT users, including free-tier, but remain undocumented in official OpenAI release notes, according to an unnamed source. <a href="https://app.daily.dev/posts/7V1v7MpYn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>React server components hit with DoS vulnerabilities: Multiple high-severity denial of service vulnerabilities (CVE-2026-23864, CVSS 7.5) have been found in React Server Components, affecting React 19.0.x through 19.2.x and Next.js versions 13.x through 16.x. These flaws can be exploited via crafted HTTP requests, potentially leading to server crashes or excessive CPU usage. Vercel has deployed WAF mitigations, but immediate upgrades to patched versions (React 19.0.4+, 19.1.5+, 19.2.4+) are recommended to fully address the issue. <a href="https://app.daily.dev/posts/igYAcLEa7" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li>Docker sandboxes isolates AI coding agents with microVMs: Docker Sandboxes now uses microVM-based isolation for AI coding agents like Claude Code and Gemini CLI, allowing them to install packages and modify files without affecting 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>AI-assisted code generation impacts open-source sustainability: A research paper suggests that ‘vibe coding’ with AI agents, which assemble software from open-source repositories without direct maintainer interaction, is leading to a decline in traditional user feedback and maintainer incentives, threatening open-source quality and sustainability. <a href="https://app.daily.dev/posts/YRC8HFFhU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>OpenAI launches AI for science team to accelerate research: OpenAI has formed an AI for Science team, led by Kevin Weil, to help scientists leverage GPT-5 for brainstorming, sourcing references, and data analysis, aiming to accelerate research despite persistent issues with model errors and hallucinations. <a href="https://app.daily.dev/posts/l0OCyxW2P" 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, offering over 10 petaFLOPS in FP4 precision and 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>Google launches optimized mobile/edge LLM for function calling: Google released FunctionGemma, a 270M parameter variant of the Gemma 3 model, specifically optimized for efficient function calling on mobile and edge hardware, enabling AI agents to translate natural language into structured API calls for on-device task automation. <a href="https://app.daily.dev/posts/DFyxBQQFb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Uncloud Kubernetes alternative gains momentum for small teams: Uncloud, an open-source deployment platform using Docker, Caddy, and WireGuard, is gaining traction as a simpler Kubernetes alternative for small teams, offering single-command deployments from source to production without container registries. <a href="https://app.daily.dev/posts/FrS1fa85P" 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 new login manager dependent on systemd-logind, effectively ending FreeBSD support and signaling an increasing Linux-specific direction for the desktop environment, mirroring similar changes in GNOME 49. <a href="https://app.daily.dev/posts/satawafBY" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li>Cursor AI agents’ browser: ‘from scratch’ or from the scrap heap?: Cursor’s CEO proudly announced their AI agents built a ‘fully functional web browser’ in a week, ‘from scratch,’ generating over 3 million lines of code. Turns out, independent developers found the browser barely compiles, doesn’t work, and heavily relies on existing projects like Servo and QuickJS. So much for AI’s miraculous virgin birth of software; seems it’s more of a copy-paste artist with a flair for marketing hyperbole. <a href="https://app.daily.dev/posts/DovPZUpOP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI coding assistants: the new knowledge privatizers: Apparently, LLM-powered coding assistants are now hoarding all the good programming knowledge, training on proprietary user data instead of those quaint public forums like Stack Overflow. The author predicts this will lead to market consolidation, paywalled access to critical programming knowledge, and regionalized pricing for your AI overlord subscriptions. Because nothing says ‘progress’ like turning open knowledge into a corporate data silo. <a href="https://app.daily.dev/posts/p61M36Xco" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI-generated software tests: a false sense of security: The latest trend is using LLMs to generate software tests after the code is written. The critique? These tests are just ‘ceremony,’ providing a false sense of security because they don’t actually validate behavior, only that the code looks right. It’s like having a student grade their own homework and then claim they’re a genius. Real tests, apparently, should fail first to prove they’re doing their job. <a href="https://app.daily.dev/posts/wLM9iDorf" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>