<h2>Must know</h2>
<ul>
<li>Tailwind CSS faces AI-driven revenue collapse and layoffs: Tailwind CSS laid off 75% of its engineering staff due to an 80% revenue decline, despite record npm downloads. The company attributes this directly to AI tools, which generate Tailwind code and answer developer questions, eliminating the need to visit their documentation site—their primary channel for selling paid products like Tailwind Plus and Catalyst. This situation, reported by multiple sources including Tailwind’s founder Adam Wathan, highlights a critical challenge for open-source projects whose business models rely on documentation traffic, as AI consumes free content without compensation. <a href="https://app.daily.dev/feed-by-ids?id=3KrG8AfuZ&amp;id=d57ILPjBN&amp;id=hsznY6oi4&amp;id=mpCeVdjI4&amp;id=re8TItYuD&amp;id=u4S1Vnazf&amp;id=yhQvgBQbs&amp;id=WwALdiGFY&amp;id=tFhv6QoOp" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Critical vulnerabilities in Coolify and HPE OneView demand immediate patching: Coolify, an open-source self-hosting platform, disclosed 11 critical vulnerabilities (CVSS 10.0) allowing authenticated attackers to achieve root access and full server compromise, with over 52,000 instances exposed globally. Separately, CISA added a critical remote code execution flaw (CVE-2025-37164, CVSS 10.0) in HPE OneView management software to its Known Exploited Vulnerabilities catalog, actively exploited in the wild. Both issues, reported by The Hacker News and Bleeping Computer, require immediate patching to prevent widespread exploitation. <a href="https://app.daily.dev/feed-by-ids?id=RXFyRlNEd&amp;id=VmTOY3pRn&amp;id=ZRs47yc16" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Linux kernel security and AI code generation spark debate: Linus Torvalds rejected AI-specific policies in Linux kernel documentation, arguing that existing code review processes are sufficient for quality control, regardless of whether code is AI-assisted. This comes as research reveals 20% of Linux kernel bugs hide for over five years, challenging the perception of Linux’s inherent security. Meanwhile, AI-generated code now accounts for 34% of development work, but 70% of DevSecOps professionals report increased compliance difficulties, highlighting the tension between AI-driven productivity and security risks, as discussed by GitLab and It’s Foss. <a href="https://app.daily.dev/feed-by-ids?id=4cToJt8jS&amp;id=TBqkb6zYC&amp;id=DuxmOezZE" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li>WebAssembly continues quiet but strong production adoption: WebAssembly is widely used in production by companies like Figma and Cloudflare, primarily as a secure, portable compilation target for languages like Rust and C++, enabling safe execution of untrusted code with sub-millisecond spinup times. <a href="https://app.daily.dev/posts/9eDXPfoXc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Ruby 4.0 released with new JIT compiler and ecosystem updates: Ruby 4.0 is out, featuring ZJIT, an optional JIT compiler for higher performance, and includes updates to Prism as the default parser, RBS for type checking, and full compatibility across key libraries like Pry and ruby.wasm. <a href="https://app.daily.dev/posts/LoVcIyHKl" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Databricks and NVIDIA advance enterprise and edge AI platforms: Databricks launched Instructed Retriever, an AI architecture integrating database queries with RAG for more accurate enterprise AI responses, while NVIDIA introduced TensorRT Edge-LLM, an open-source C++ framework for high-speed LLM inference on automotive and robotics edge systems. <a href="https://app.daily.dev/feed-by-ids?id=YccZnqmwj&amp;id=1Lrxf6i15&amp;id=IYZKv2e0G&amp;id=NYhUyiiE6&amp;id=Ss6K85gTn&amp;id=vZbhC083l" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Cisco patches ISE and Snort vulnerabilities: Cisco released patches for medium-severity vulnerabilities in Identity Services Engine (ISE) and Snort 3 Detection Engine, with a public proof-of-concept exploit for the ISE flaw increasing the urgency for administrators to update. <a href="https://app.daily.dev/feed-by-ids?id=y9ijzcFQJ&amp;id=AOIh6SOOs&amp;id=gIGTza698" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>CrowdStrike acquires SGNL for identity security: CrowdStrike acquired SGNL for $740 million to bolster its Falcon platform’s identity security, focusing on continuous, context-aware authorization for human and AI agents amid a 32% rise in identity-based cyberattacks. <a href="https://app.daily.dev/feed-by-ids?id=kHk9YWnOk&amp;id=kl0LpaKDU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Malicious npm packages distribute NodeCordRAT malware: Three malicious npm packages impersonating legitimate bitcoinjs projects were downloaded over 3,400 times, deploying NodeCordRAT malware to steal credentials and cryptocurrency wallet seed phrases, highlighting critical software supply chain security gaps. <a href="https://app.daily.dev/feed-by-ids?id=90EDnfct7&amp;id=yWKZKkjtB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Linux kernel considers PIE for x86_64 for enhanced security: The Linux kernel is considering linking the relocatable x86_64 kernel as Position Independent Executable (PIE) to enable advanced security features like fine-grained KASLR, with minimal code size increase and no measurable performance impact. <a href="https://app.daily.dev/posts/oWVbWSMtS" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li>AI models leak copyrighted works and private data, but don’t worry, they’re ‘learning’: Stanford and Yale researchers found that AI models from Anthropic, Google, and xAI can reproduce large verbatim sections of copyrighted books, with some models spitting out over 70% of ‘Harry Potter’ without even trying. Meanwhile, OpenAI’s ChatGPT is vulnerable to a new ‘ZombieAgent’ attack that exfiltrates user data, bypassing previous defenses. Apparently, these AI systems are so advanced they can plagiarize and leak your data with impressive efficiency, but don’t worry, the industry assures us they’re just ‘reactively patching’ these minor oversights. <a href="https://app.daily.dev/feed-by-ids?id=RZ3NYDnWq&amp;id=pnbMxpKrW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI-driven e-commerce: because what we really need is more personalized ads from a six-person startup: Spangle, an AI e-commerce startup with a whopping six employees, just secured $15 million at a $100 million valuation by promising ‘personalized, real-time shopping experiences’ that dynamically generate site content. Their clients report 50% revenue increases and doubled ad spend return. So, a tiny team is making big bucks by making your shopping experience even more tailored, while Shopify rolls out its own AI assistant, Sidekick. It’s almost like the entire industry is racing to make sure you can’t escape targeted advertising, all while claiming ‘innovation.’ <a href="https://app.daily.dev/feed-by-ids?id=ntZfQKCbY&amp;id=sXlQ6vcGQ&amp;id=ABax1QAt1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Enterprise AI workflows: automating the obvious, still years from critical tasks: Dynatrace launched ‘agentic workflows’ using Davis CoPilot to automate monitoring and analysis, claiming to handle incidents and security threats by autonomously scaling resources. They’re focusing on ‘simple, repetitive tasks’ like AIOps anomaly detection and cloud cost management. Experts, however, are quick to point out that fully autonomous workflows for non-critical background processes are expected by late 2026, but for mission-critical infrastructure, we’re still 5-10 years out. So, AI can manage your cloud bill, but don’t expect it to run your core systems without human babysitting anytime soon. Progress, I guess? <a href="https://app.daily.dev/feed-by-ids?id=LH3z3eMJi&amp;id=gQn7Cgin4" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>