<h2>Must know</h2>
<ul>
<li><strong>AI-generated code is breaking open source</strong>: Open-source maintainers are getting swamped by low-quality, AI-generated pull requests, leading projects like TL Draw, Curl, and Zig to restrict contributions. This ‘AI slop’ looks good on the surface but creates technical debt and drains maintainer resources, as reported by multiple sources including Frontend Masters. The ease of AI tools means contributors bypass learning project architecture, threatening the sustainability of open-source projects that rely on web traffic for revenue and community engagement. Tailwind Labs, for example, saw a 40% drop in documentation visits and laid off 75% of its staff, directly attributing it to AI-assisted coding. <a href="https://app.daily.dev/feed-by-ids?id=9Rmqfgu76&amp;id=3KjVRqoNX&amp;id=PBAzOX1yL&amp;id=YRC8HFFhU&amp;id=YH0ajyAf5&amp;id=aDodkq3hd&amp;id=vz4g5mSDD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>React Server Components have a DoS vulnerability</strong>: Multiple high-severity denial-of-service vulnerabilities (CVE-2026-23864) have been found in React Server Components, affecting versions 19.0.x through 19.2.x. These flaws can be triggered by specially crafted HTTP requests to Server Function endpoints, potentially causing server crashes or excessive CPU usage. Next.js versions 13.x through 16.x are also affected. Vercel has deployed WAF mitigations, but immediate upgrades to patched React and Next.js versions are strongly advised to prevent exploitation. <a href="https://app.daily.dev/feed-by-ids?id=igYAcLEa7&amp;id=igYACLEa7&amp;id=91xiEmcVe" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents are still pretty dumb at complex tasks</strong>: A new benchmark, Apex-Agents, designed to test AI models on white-collar professional tasks like consulting and investment banking, found that even leading models like Gemini 3 Flash and GPT-5.2 achieved a dismal 24% accuracy. The study, conducted by Mercor, highlights significant gaps in current AI agent capabilities for complex, multi-domain reasoning and tool usage. While AI performance is improving year-over-year, don’t expect your AI assistant to replace your knowledge workers just yet. <a href="https://app.daily.dev/posts/OsmrZ9UpM" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Microsoft Word gets an AI agent mode</strong>: Microsoft Word now features an ‘Agent Mode’ for Copilot AI, allowing it to draft, refine, and format documents conversationally, leveraging ‘Work IQ’ to access organizational and personal data. <a href="https://app.daily.dev/posts/4kNrEEPPk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Alibaba launches Qwen3-Max-Thinking for enterprises</strong>: Alibaba Cloud has unveiled Qwen3-Max-Thinking, an enterprise-focused AI reasoning model that claims to rival GPT-5.2-Thinking, Claude-Opus-4.5, and Gemini 3 Pro on 19 benchmarks, targeting enterprises seeking alternatives to US-based models. <a href="https://app.daily.dev/posts/NV9ulemJA" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered phishing is getting more sophisticated</strong>: Security researchers have demonstrated a new attack using LLM APIs to generate unique, polymorphic JavaScript phishing payloads dynamically at runtime, bypassing traditional network and signature-based detections. <a href="https://app.daily.dev/posts/6x4IdBLjT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-driven malware VoidLink emerges</strong>: Check Point Research has identified VoidLink, the first fully AI-generated advanced malware, developed by a single actor using TRAE IDE and AI agents, producing over 88,000 lines of code in under a week. <a href="https://app.daily.dev/posts/nLTC7GTyu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New Linux kernel patch for Xeon wakeup latency</strong>: A proposed Linux kernel patch aims to reduce CPU wakeup latency on Intel Xeon Sapphire Rapids platforms from 151μs to 30μs by adjusting idle state prediction logic, with no reported power or performance regressions on older Xeons. <a href="https://app.daily.dev/posts/NWx6tDyeo" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-driven UI component library launched</strong>: SpicyUI has launched a free, AI-powered UI component library for React and Next.js developers, featuring AI-generated components and a community feed for sharing and remixing. <a href="https://app.daily.dev/posts/2aQsio3bN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered price tracking app for e-commerce</strong>: PriceGhost is a new self-hosted application that monitors e-commerce product prices using AI-powered extraction (Claude, GPT, Ollama fallback) and provides price drop and restock notifications. <a href="https://app.daily.dev/posts/51ovFH0wZ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>OpenAI’s $500 billion Stargate datacenter: because ‘community plans’ fix everything</strong>: OpenAI has unveiled ‘community plans’ for its $500 billion Stargate datacenter project, promising to fund its own energy infrastructure and use closed-loop cooling to address local concerns about electricity and water. This comes after 20 US datacenter projects were blocked or delayed in Q2 2025 due to widespread opposition. OpenAI is clearly hoping that a few workforce development programs and a ‘we care’ attitude will make everyone forget about the environmental impact and the fact that each facility will only employ about 100 people. It’s almost like they think a half-trillion-dollar project can be greenwashed with a press release. <a href="https://app.daily.dev/feed-by-ids?id=O0VoWxWDb&amp;id=O0VoWxWDb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Apple’s AI pin: because Humane AI’s failure wasn’t a clear enough signal</strong>: Apple is reportedly developing an AI-powered wearable ‘AI pin’ with dual cameras and three microphones, aiming for a 2027 launch and 20 million units shipped. This accelerated development supposedly comes after OpenAI announced its own AI hardware. Apparently, the spectacular failure of Humane AI’s similar wearable pin, which shut down within two years, wasn’t enough to deter Apple. Because if at first you don’t succeed, just throw more money and a fruit logo at it, right? <a href="https://app.daily.dev/posts/fgglWe8Qn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Tech CEOs predict six-figure salaries for electricians: because AI needs more power, not people</strong>: Nvidia’s Jensen Huang and Palantir’s Alex Karp are touting AI’s infrastructure buildout as the ‘largest construction project in history,’ promising six-figure salaries for electricians and construction workers. Microsoft’s Satya Nadella, ever the pragmatist, cautiously agrees, but only if AI delivers ‘quantifiable benefits’ in healthcare and education. Meanwhile, research estimates 6-10% of US jobs could be displaced by AI by 2030. It’s a classic tech executive move: distract from job displacement fears by hyping up a few high-paying blue-collar jobs, as if a few thousand electricians will absorb millions of displaced workers. The ‘AI boom’ is apparently powered by both silicon and wishful thinking. <a href="https://app.daily.dev/posts/t53FyhsIC" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
