<h2>Must know</h2>
<ul>
<li><strong>AI agents are getting serious, but so are the risks</strong>: AI agents are moving beyond chatbots, with tools like StepFun AI’s Step-DeepResearch and NeuBird’s Hawkeye automating complex tasks from scientific research to incident response. However, this autonomy comes with significant risks: Gartner warns of ‘model collapse’ from AI-generated data contamination, and a production incident revealed a single compromised agent could crash 50 others due to inadequate trust infrastructure. Meanwhile, Docker Sandboxes is tackling security by isolating AI coding agents in microVMs, and OpenAI’s ChatGPT now allows Bash commands and package installation in its sandbox, expanding capabilities but also potential attack surfaces. <a href="https://app.daily.dev/feed-by-ids?id=QxbydGQJy&amp;id=Skhkgwj1Z&amp;id=qF5kIahRI&amp;id=uTj3yEY4b&amp;id=EGTSOLShb&amp;id=cIiWYuVU9&amp;id=mMOsDNB8r&amp;id=6YdOupFK7&amp;id=7V1v7MpYn&amp;id=cOrYC12OX" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft and Google push new AI hardware and platforms</strong>: Microsoft has unveiled the Maia 200 AI inference accelerator, built on TSMC’s 3nm process, promising over 10 petaFLOPS in FP4 and 30% better performance per dollar for tasks like GPT-5.2 models and Microsoft 365 Copilot. Not to be outdone, Google launched Gemini for Government, an AI platform authorized at FedRAMP High, already adopted by the Department of Defense and FDA. This signals a major push for AI integration into critical infrastructure, with both companies vying for dominance in the AI hardware and platform space. <a href="https://app.daily.dev/feed-by-ids?id=HBgJElvHn&amp;id=Fm4JFG1tC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Open source sustainability under threat from AI</strong>: The open-source ecosystem is facing an existential crisis, with projects like Tailwind CSS reporting an 80% drop in revenue and 75% staff layoffs, despite massive download numbers. The culprit? Developers increasingly using AI tools, bypassing traditional documentation and support channels that once generated revenue. A research paper on ‘vibe coding’ suggests this trend reduces maintainer incentives, threatening OSS quality. Frontend Masters is stepping up with a $50,000 commitment to open source, but the industry needs to figure out new compensation models to keep critical infrastructure alive. <a href="https://app.daily.dev/feed-by-ids?id=3KjVRqoNX&amp;id=YRC8HFFhU" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Nvidia invests big in AI compute capacity</strong>: Nvidia is pouring $2 billion into CoreWeave to accelerate the construction of over 5 gigawatts of AI computing infrastructure by 2030, integrating Nvidia’s Rubin chip architecture and Bluefield storage solutions. <a href="https://app.daily.dev/posts/suyuzcb2U" target="_blank" rel="noopener nofollow">Read more</a></li>
<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 led to its disabling in the rc7 release candidate. <a href="https://app.daily.dev/posts/85ivk605B" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AMDGPU Linux driver gets HDMI VRR and ALLM</strong>: AMDGPU Linux kernel driver patches now introduce HDMI Variable Rate Refresh (VRR) and Auto Low Latency Mode (ALLM) for gaming on HDMI displays, aiming for parity with FreeSync capabilities. <a href="https://app.daily.dev/posts/bsF1QAXcQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust core team member builds AI-assisted systems language</strong>: Steve Klabnik, a Rust core team member, has unveiled Rue, a new memory-safe systems programming language developed with substantial assistance from Anthropic’s Claude AI, exploring how AI can enable single developers to deliver complex infrastructure software. <a href="https://app.daily.dev/posts/scnD6visU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Uncloud gains traction as Kubernetes alternative</strong>: Uncloud, an open-source deployment platform built on Docker, Caddy, and WireGuard, is gaining momentum among small teams as a simpler Kubernetes alternative, offering single-command deployments from source to production. <a href="https://app.daily.dev/posts/FrS1fa85P" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ethereum Foundation prioritizes post-quantum security</strong>: The Ethereum Foundation has made post-quantum security a top strategic priority, establishing a dedicated research and engineering team and offering a $2 million research prize pool for hash-based cryptography. <a href="https://app.daily.dev/posts/Ub6V8IYS8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cisco integrates eBPF for modern firewall architecture</strong>: Cisco is implementing eBPF technology, acquired through Isovalent, into its enterprise networking hardware and software, including Tetragon and Hypershield, to move beyond traditional centralized firewalls to device- and application-level security. <a href="https://app.daily.dev/posts/VWEt8pf7l" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Cursor AI’s ‘from scratch’ browser: more like ‘from existing projects and failing’</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 relies heavily on existing projects like Servo and QuickJS. So, ‘from scratch’ apparently means ‘cobbled together with AI, mostly from other people’s work, and it’s still broken.’ Impressive marketing, less impressive reality. <a href="https://app.daily.dev/posts/DovPZUpOP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI coding assistants: the ultimate productivity paradox</strong>: A 2025 study claims developers using AI assistants took 19% longer to complete tasks, despite feeling faster. Apparently, AI-generated code leads to ‘architectural degradation’ and increased code review times by 91%, with 45% containing security vulnerabilities. So, we’re faster at writing bad code that takes longer to fix and review, and is less secure. Sounds like a real win for ‘productivity.’ <a href="https://app.daily.dev/feed-by-ids?id=pYjmmMIkn&amp;id=MVxk3C8r1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ChatGPT and Claude: now citing AI-generated ‘facts’ from Grokipedia</strong>: It seems our leading LLMs, ChatGPT (GPT-5.2) and Anthropic’s Claude, are now citing Grokipedia—Elon Musk’s AI-generated, conservative-leaning, and notoriously inaccurate encyclopedia—as a source. Especially for ‘obscure topics.’ So, not only do we have to worry about AI hallucinating, but now it’s confidently pulling ‘facts’ from other AI’s questionable fan fiction. Transparency? Source validation? Who needs it when you have Grokipedia! <a href="https://app.daily.dev/posts/SDpas3mop" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
