<h2>Must know</h2>
<ul>
<li><strong>AI is gutting open-source business models</strong>: Tailwind CSS laid off 75% of its engineering team due to an 80% revenue drop, directly attributing the decline to AI tools. Developers are using LLMs to generate code, bypassing official documentation—the primary channel for upselling paid products like Tailwind Plus and Catalyst. This highlights a critical vulnerability in open-source monetization, where increased usage doesn’t translate to revenue when AI consumes content without compensation, as reported by multiple sources including Tailwind maintainers and tech industry reporters. Google AI Studio has since announced sponsorship, but the long-term impact on open-source sustainability remains a major concern. <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>Critical vulnerabilities in HPE OneView and Coolify demand immediate action</strong>: CISA has flagged a critical remote code execution vulnerability (CVE-2025-37164, CVSS 10.0) in HPE OneView, actively exploited in the wild, allowing full system control. HPE released a hotfix, and immediate patching is crucial due to OneView’s administrative privileges. Separately, Coolify, an open-source self-hosting platform, disclosed 11 critical flaws (CVSS 10.0) enabling root command execution and full server compromise on its 52,890 publicly exposed instances. While no active exploitation is confirmed for Coolify, immediate upgrades to patched versions are advised, as detailed by The Hacker News and CISA advisories. <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>
<li><strong>AI models are leaking copyrighted data and user information</strong>: Researchers from Stanford and Yale found that major AI models like Anthropic’s Claude 3.7 Sonnet and Google’s Gemini 2.5 Pro can reproduce large verbatim sections of copyrighted books, even without ‘jailbreaking.’ This direct output weakens fair use claims and could bolster ongoing copyright lawsuits. Concurrently, OpenAI’s ChatGPT is vulnerable to a new ‘ZombieAgent’ attack that bypasses existing defenses to exfiltrate user data from its servers. These incidents highlight persistent security and intellectual property risks in commercial AI systems, as reported by academic research teams. <a href="https://app.daily.dev/feed-by-ids?id=RZ3NYDnWq&amp;id=pnbMxpKrW" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Linus Torvalds rejects AI-specific policies for Linux kernel</strong>: Linus Torvalds dismissed proposals to require disclosure of AI-assisted contributions to the Linux kernel, stating 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>Google and <a href="http://Character.AI" target="_blank" rel="noopener nofollow">Character.AI</a> settle AI chatbot harm lawsuits</strong>: Google and <a href="http://Character.AI" target="_blank" rel="noopener nofollow">Character.AI</a> have reached undisclosed settlements in multiple lawsuits alleging emotional harm to minors from AI chatbots, including a case linked to a teenager’s suicide, following Google’s $2.7 billion licensing deal with <a href="http://Character.AI" target="_blank" rel="noopener nofollow">Character.AI</a>. <a href="https://app.daily.dev/feed-by-ids?id=G2f0bA0Zl&amp;id=jKMfXmxBn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>CrowdStrike acquires SGNL for identity security</strong>: 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, addressing 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><strong>Ruby 4.0 released with new JIT compiler and ecosystem updates</strong>: Ruby 4.0 is out, featuring ZJIT, an optional JIT compiler, Prism as the default parser, and updated ecosystem tools like RBS 3.10 and ruby.wasm 2.8, ensuring broad compatibility and performance improvements. <a href="https://app.daily.dev/posts/LoVcIyHKl" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks introduces Instructed Retriever for enterprise AI accuracy</strong>: Databricks unveiled Instructed Retriever, an AI architecture that integrates database queries with RAG similarity search, parsing user instructions before retrieval to filter documents based on metadata and business rules, aiming for more accurate enterprise AI responses. <a href="https://app.daily.dev/posts/YccZnqmwj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA optimizes Linux memory stats printing</strong>: An NVIDIA engineer optimized Linux kernel memory statistics printing, reducing system time by 11% when dumping memory stats by replacing printf-based functions with lightweight helpers in the memory resource controller. <a href="https://app.daily.dev/posts/ncmgboYR1" 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 using Google Cloud’s DORA framework, building a multi-million dollar sandbox to reduce User Acceptance Testing cycles and targeting 100% DORA metrics adoption across engineering teams for faster regulatory software updates. <a href="https://app.daily.dev/feed-by-ids?id=ABax1QAt1&amp;id=ntZfQKCbY" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI will make you a ‘manager’ of code, not a coder</strong>: Simon Willison, on the Oxide and Friends podcast, predicts that manually typing code will be obsolete within six years, likening it to punch cards. Apparently, software engineering will persist, but it’ll be all about ‘management skills’ for coordinating AI agents. So, instead of writing code, you’ll be managing robots that write code, which sounds suspiciously like more management and less actual engineering. <a href="https://app.daily.dev/feed-by-ids?id=QaeYUARam&amp;id=cLDAM9Yxb&amp;id=tmcYDkbnT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google’s Gemini 3: the ‘unified’ AI platform that will solve all your problems</strong>: Google’s Gemini 3 is here, promising to ‘redefine multimodal and agentic computing’ by handling massive inputs and offering ‘Deep Think’ mode for advanced reasoning. It’s supposed to replace all your ‘piecemeal’ AI systems with a single API. Because, you know, fragmentation is bad, and a single vendor solution is always the answer. Just don’t ask about the ‘thorough governance, testing, and monitoring frameworks’ you’ll need to deploy it effectively. <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>AI-generated malware: the next big thing in ‘security theater’</strong>: Palo Alto Networks’ Unit 42 reports a rise in cybercriminals using AI to create malware, complete with ‘watermarked code’ and AI-specific typos. Apparently, some of this AI-generated malware is even ‘ineffective or non-functional,’ which they’ve dubbed ‘security theater.’ So, we’re not just getting bad code from AI, we’re getting bad <em>malware</em> from AI. What a time to be alive. <a href="https://app.daily.dev/feed-by-ids?id=keYljc5pY&amp;id=ntbZHxeN3" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
