<h2>Must know</h2>
<ul>
<li><strong>AI is eating open-source revenue models, leading to layoffs</strong>: Tailwind CSS laid off 75% of its engineering staff due to an 80% revenue decline, directly attributed to AI tools answering developer questions and generating code, bypassing their documentation-based monetization. This highlights a critical vulnerability for open-source projects whose business models rely on website traffic and engagement, as AI consumes content without compensation. Google AI Studio has stepped in with sponsorship, but the broader issue of AI’s impact on open-source sustainability remains a significant concern, as reported by multiple tech outlets. <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><strong>AI models are leaking copyrighted data and user info</strong>: New research from Stanford and Yale reveals that major AI models like Anthropic’s Claude 3.7 Sonnet and Google’s Gemini 2.5 Pro can reproduce large portions of copyrighted books verbatim, even without ‘jailbreaking.’ This directly challenges claims of ‘transformativeness’ in ongoing copyright lawsuits. Separately, OpenAI’s ChatGPT is vulnerable to a new ‘ZombieAgent’ attack that can exfiltrate user data, bypassing previous defenses. These incidents, detailed in academic reports, underscore persistent security and intellectual property risks in commercial AI systems. <a href="https://app.daily.dev/feed-by-ids?id=RZ3NYDnWq&amp;id=pnbMxpKrW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical vulnerabilities in HPE OneView and Coolify demand immediate attention</strong>: CISA has flagged a critical remote code execution vulnerability (CVE-2025-37164) in HPE OneView, rated CVSS 10.0, which is actively being exploited and allows full system control. HPE has released a hotfix, urging immediate patching. Separately, Coolify, an open-source self-hosting platform, disclosed 11 critical vulnerabilities, also rated CVSS 10.0, allowing root command execution and container escapes on over 52,000 publicly exposed instances. While no active exploitation is confirmed for Coolify, immediate patching is advised for both, as reported by Bleeping Computer and Coolify’s own disclosure. <a href="https://app.daily.dev/feed-by-ids?id=VmTOY3pRn&amp;id=RXFyRlNEd&amp;id=ZRs47yc16" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>NVIDIA and Google push multimodal and edge AI boundaries</strong>: NVIDIA launched TensorRT Edge-LLM, an open-source C++ framework for high-speed LLM inference on automotive and robotics systems, while Google introduced Gemini 3, capable of handling massive multimodal inputs and offering advanced reasoning for complex enterprise tasks. <a href="https://app.daily.dev/feed-by-ids?id=1Lrxf6i15&amp;id=IYZKv2e0G&amp;id=NYhUyiiE6&amp;id=Ss6K85gTn&amp;id=vZbhC083l" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linus Torvalds rejects AI-specific policies for Linux kernel</strong>: Linus Torvalds has dismissed proposals to add AI-specific disclosure requirements to the Linux kernel documentation, arguing that AI is ‘just a tool’ and existing code review processes are sufficient to handle low-quality submissions, regardless of their origin. <a href="https://app.daily.dev/posts/4cToJt8jS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>CrowdStrike acquires SGNL for $740M to boost identity security</strong>: CrowdStrike acquired SGNL for $740 million to enhance its Falcon platform with continuous, context-aware identity authorization, addressing the rise in identity-based cyberattacks and the need to secure both human and AI/machine agents. <a href="https://app.daily.dev/feed-by-ids?id=kHk9YWnOk&amp;id=kl0LpaKDU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ruby 4.0 released with new JIT compiler and ecosystem updates</strong>: Ruby 4.0 is out, featuring ZJIT, a new optional JIT compiler, and solidifying Prism as the default parser, alongside updates to RBS, Pry, sqlite3-ruby, and ruby.wasm, ensuring ecosystem readiness. <a href="https://app.daily.dev/posts/LoVcIyHKl" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ustwo Games shifts focus from mobile to PC development</strong>: Ustwo Games, known for Monument Valley, is pivoting its primary development focus to PC, citing limitations of a ‘mobile-first’ approach for broader platform success and aiming to build more direct player relationships. <a href="https://app.daily.dev/posts/SALrx3zyi" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>FINRA adopts DORA framework for DevOps transformation</strong>: FINRA is modernizing its software development process by adopting Google Cloud’s DORA framework, aiming to reduce long User Acceptance Testing cycles and standardize improvement efforts across 50% of its engineering teams. <a href="https://app.daily.dev/posts/ABax1QAt1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes sustainability and cloud-native trends emerge</strong>: kube-green, a Kubernetes operator, automates pod shutdowns to reduce resource consumption and CO2 emissions, reflecting a broader industry shift towards sustainability-focused open-source adoption and efficient cloud management. <a href="https://app.daily.dev/feed-by-ids?id=CQDSjNmmR&amp;id=uWdRVw1vL" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-driven ‘vibe coding’ is here to save us all (or just make more bugs)</strong>: Apparently, ‘vibe coding’—where AI generates entire programs with minimal human input—is the ‘most significant trend of the year,’ according to Norwegian tech outlets. While it promises ‘rapid productivity,’ security experts are already seeing breaches from missing authentication and prompt injection attacks. So, while management pushes for faster AI-generated code, developers are left to clean up the ‘security theater’ and accidental database deletions. What could go wrong when we let the vibes dictate our codebase? <a href="https://app.daily.dev/feed-by-ids?id=CSSbkecce&amp;id=mpCeVdjI4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Enterprise AI workflows: fully autonomous in 5-10 years, but totally ready now!</strong>: Dynatrace is launching ‘agentic workflows’ to automate monitoring and response with generative AI, promising ‘human-like judgment.’ While industry experts cautiously predict full autonomy for mission-critical infrastructure is 5-10 years away, vendors are already marketing ‘end-to-end workflow automation’ as a present-day reality. It seems the ‘critical’ part of ‘mission-critical’ is just a suggestion when there’s a new AI buzzword to sell. <a href="https://app.daily.dev/feed-by-ids?id=LH3z3eMJi&amp;id=gQn7Cgin4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks’ Instructed Retriever: RAG, but with more instructions (and hype)</strong>: Databricks unveiled ‘Instructed Retriever,’ claiming it delivers ‘more accurate AI-generated answers’ by parsing user instructions <em>before</em> retrieval, unlike standard RAG. Analysts, ever so cautiously, note it ‘more closely aligns retrieval with business needs.’ So, it’s RAG, but with extra steps and a new name, promising accuracy without any actual benchmarking evidence. Just make sure your ‘data maturity’ and ‘metadata quality’ are top-notch, or this ‘superior’ architecture might just be another expensive toy. <a href="https://app.daily.dev/posts/YccZnqmwj" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
