<h2>Must know</h2>
<ul>
<li><strong>AI coding agents are changing everything, but not how you think</strong>: Anthropic’s CEO predicts AI models could handle most software engineering tasks in 6-12 months, with some engineers already using AI to write code, leaving humans to edit. However, a new report highlights that the real bottleneck for AI coding tools is ‘context transfer’—getting tacit human knowledge into AI systems. This means developers will shift from coding to architecting, reviewing, and managing AI agents, as noted by Jeffrey Wei, who saw a 40% staff reduction in his ed-tech platform due to AI coding agents. The future isn’t about AI replacing developers, but about developers evolving to manage AI, as discussed by Sumit Saha on freeCodeCamp. <a href="https://app.daily.dev/feed-by-ids?id=0fyNmKnQE&amp;id=DczaKkmGw&amp;id=KfJ8s7nDm&amp;id=Q8EgPml80&amp;id=7YXm1V9vt&amp;id=XUOoVy4Ig&amp;id=dSFznJ0W1&amp;id=TFZmmg40D" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft’s BitLocker keys and SCOM deprecation raise privacy and audit concerns</strong>: Microsoft provided the FBI with BitLocker recovery keys in a fraud investigation, marking the first publicly reported instance of such cooperation. This highlights privacy risks, especially since BitLocker keys are uploaded to Microsoft’s cloud by default. Separately, Microsoft is deprecating SCOM Management Packs, forcing organizations to migrate to Azure-based monitoring. This move, while offering unified monitoring, raises concerns among analysts about increased exposure to Microsoft’s license compliance audits, as registering on-premises servers with Azure gives Microsoft greater visibility into customer environments. <a href="https://app.daily.dev/feed-by-ids?id=6Aq315WBk&amp;id=Lyoqg9qSt" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New AI model security flaws and supply chain attacks demand attention</strong>: A critical remote code execution vulnerability (CVSS 9.8) in Trend Micro Apex Central allows attackers to load arbitrary DLLs with SYSTEM privileges, requiring no shellcode. Meanwhile, a security researcher demonstrated a zero-click exploit targeting AI coding agents in VS Code and Cursor via malicious tasks.json files, capable of exfiltrating sensitive data like API keys. CISA also added eslint-config-prettier to its KEV catalog due to a supply chain attack, where malicious code was embedded after maintainer credentials were phished, underscoring the growing risks in AI and software supply chains. <a href="https://app.daily.dev/feed-by-ids?id=bkT31Gnrl&amp;id=YV7AnAdiX&amp;id=io0G9RHpN" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AWS launches M4 Max Mac instances for Apple app development</strong>: AWS has released M4 Max Mac instances for EC2, utilizing Apple M4 Max chips to offer up to 25% better application build performance for iOS, macOS, and other Apple platform apps. <a href="https://app.daily.dev/posts/DEbhYteDE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vercel’s json-render aims to automate UI generation with AI</strong>: Vercel launched json-render, an open-source tool that uses AI to generate user interfaces from natural language descriptions, allowing developers to specify UI components and data bindings as guardrails. <a href="https://app.daily.dev/posts/5RKyfi3u8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s Agent Skills specification gains traction</strong>: Anthropic’s Agent Skills, a new specification for AI coding agents, is gaining adoption with Claude Code and Cursor, allowing structured guidance and progressive discovery for AI agents. <a href="https://app.daily.dev/feed-by-ids?id=s7kFlYuVy&amp;id=lvPOiATWV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub Copilot CLI integrates organizational context with Work IQ</strong>: GitHub Copilot CLI now incorporates Work IQ, enabling AI agents to access organizational data from documents, emails, and SharePoint, allowing them to compare code against company specifications and reconstruct undocumented architecture. <a href="https://app.daily.dev/posts/AnhMtsEYf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MySQL community debates future amid Oracle governance concerns</strong>: MySQL community developers are discussing potential forks or increased pressure on Oracle due to perceived declines in code commits and innovation following recent layoffs in Oracle’s MySQL core team. <a href="https://app.daily.dev/posts/l2bePOl0A" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust 1.93 improves DNS resolver for static Linux binaries</strong>: Rust 1.93 upgraded its bundled musl library, bringing significant DNS resolver improvements that enhance reliability for static Linux binaries, especially when handling large DNS records. <a href="https://app.daily.dev/posts/wF6DW7Hp4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 6.19 NEXT_BUDDY scheduler disabled due to performance regressions</strong>: The Linux 6.19 kernel’s re-implemented NEXT_BUDDY scheduler has been disabled by default after performance testing revealed regressions in workloads like MySQL and Nginx HTTPS server. <a href="https://app.daily.dev/posts/iDlOb4czh" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic CEO predicts AI will replace software engineers in 6-12 months</strong>: Dario Amodei, CEO of Anthropic, boldly declared at Davos that AI models could handle most software engineering tasks within 6-12 months, with humans relegated to editing. He admits uncertainty in his timeline, but hey, why let a little thing like ‘uncertainty’ get in the way of a good headline? Meanwhile, the rest of us are still trying to get AI to write a coherent unit test, let alone an entire codebase. <a href="https://app.daily.dev/posts/0fyNmKnQE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code is replacing libraries, apparently</strong>: A new trend suggests developers are using AI to generate custom code instead of relying on third-party libraries like Tkumi and fast-float. The claim is that this reduces supply chain risks and dependency bloat. Because, you know, a black-box AI generating code is definitely more ‘maintainable’ and ‘secure’ than a well-vetted, open-source library. What could possibly go wrong? <a href="https://app.daily.dev/posts/kqbTJ9NPQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI is now an autonomous corn farm manager</strong>: Claude Code, an AI system, has apparently been employed as an ‘autonomous farm manager’ for corn farming, aggregating sensor data and weather forecasts to direct human operators. It even manages supplier relationships and logs decisions for ‘transparency.’ Because nothing says ‘cutting-edge AI’ like telling a farmer when to plant corn, a task that has been successfully managed by humans for millennia without a single line of Python. <a href="https://app.daily.dev/posts/bzSye8m41" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
