<h2>Must know</h2>
<ul>
<li><strong>Supply chain attacks target Notepad++ and eScan antivirus</strong>: State-sponsored hackers, suspected to be Chinese, compromised the Notepad++ update infrastructure from June to December 2025, delivering backdoored versions to specific targets. This incident mirrors the SolarWinds attack, highlighting persistent supply chain vulnerabilities. Separately, eScan antivirus also experienced a supply-chain compromise, as reported in a February 2026 cyber threat intelligence update. These events underscore the critical need for robust software supply chain security and vigilance. <a href="https://app.daily.dev/feed-by-ids?id=diIJMavHg&amp;id=HoGCUtbn9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft’s AI code contributions linked to quality control issues</strong>: Microsoft’s Windows team is facing a troubling trend of increased out-of-band emergency patches, with nearly every Patch Tuesday requiring subsequent fixes. This coincides with Microsoft’s claims that over 30% of its code now comes from AI contributions and recent workforce reductions. The situation raises significant questions about the quality control processes for AI-generated code and its impact on system stability, leaving system administrators in a bind between critical security updates and potential productivity losses. <a href="https://app.daily.dev/posts/H1aWwdchu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Raspberry Pi prices jump due to memory costs</strong>: Raspberry Pi announced price increases of up to $60 on several single-board computer models, including the Pi 4 and Pi 5, due to soaring LPDDR4 memory costs. This marks a repeat of last year’s adjustments, driven by high demand from the AI infrastructure sector. While models with 1GB RAM remain unaffected, this trend indicates ongoing supply chain pressures and the broader impact of AI’s resource demands on the embedded systems market. <a href="https://app.daily.dev/posts/Rk4TtZc2A" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>OpenAI launches macOS app for Codex AI agent</strong>: OpenAI released a macOS desktop app for Codex, its AI coding agent, shifting focus to managing multiple parallel AI agents for complex, long-running tasks, integrating with GitHub and supporting Git worktrees. <a href="https://app.daily.dev/posts/c2BjVZb6r" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Firefox 148 to introduce granular AI controls</strong>: Firefox 148, set for release in late February 2026, will include a new ‘AI Controls’ panel, allowing users to disable generative AI features at a granular level or entirely with a single toggle, addressing previous user backlash. <a href="https://app.daily.dev/posts/K3PNUlsQj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Snowflake and OpenAI partner for enterprise AI solutions</strong>: Snowflake and OpenAI announced a $200 million partnership to integrate OpenAI models directly into Snowflake Cortex AI, enabling natural language querying and processing of enterprise data, bypassing previous Azure-mediated access. <a href="https://app.daily.dev/posts/QNlzvTu0k" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux From Scratch drops SysVinit support for systemd-only future</strong>: Linux From Scratch (LFS) will become systemd-only starting with version 13.0, due to overwhelming volunteer workload managing two init systems and increasing upstream dependencies on systemd from major desktop environments. <a href="https://app.daily.dev/posts/ohWaYbFxO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New developer tools for AI integration and UI development emerge</strong>: New tools like <a href="http://ui.sh" target="_blank" rel="noopener nofollow">ui.sh</a> for AI-assisted UI development, Vibe for lightweight Linux VMs on ARM Macs for LLM agents, and Reposit.bot for sharing AI conversation solutions are launching, indicating a growing ecosystem for AI-driven developer workflows. <a href="https://app.daily.dev/feed-by-ids?id=FBZjjQkX5&amp;id=XJySsKWNc&amp;id=YD299xQxI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS enhances services with new features and cost allocation</strong>: AWS HealthImaging now supports JPEG XL for medical images, Amazon Connect introduced APIs for automated contact center testing, and AWS Network Firewall offers flexible cost allocation via Transit Gateway attachments in GovCloud, enhancing cloud capabilities and management. <a href="https://app.daily.dev/feed-by-ids?id=AlDvfa64l&amp;id=IXQflAT6d&amp;id=LWsmZj1tc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Git 2.53 released with Rust support, 1GB THPs for Linux kernel</strong>: Git 2.53 was released, default-enabling Rust support in its build systems, while experimental Linux kernel patches for 1GB PUD-level Transparent Huge Pages show faster memory access times but slower page fault allocation. <a href="https://app.daily.dev/feed-by-ids?id=st4SfbY5r&amp;id=tPbz1g571&amp;id=xWwOyitRt" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Cloudflare’s ‘production-grade’ matrix homeserver was just AI-generated vaporware</strong>: Cloudflare proudly announced a ‘production-grade’ Matrix homeserver built on Workers, only for Matrix co-founder Matthew Hodgson to point out it lacked critical features like federation, proper authentication, and encryption. Turns out, it was mostly unreviewed, AI-generated code with ‘TODO’ comments in critical sections. They later added a disclaimer, reclassifying it as a ‘proof of concept.’ So much for AI-assisted infrastructure, eh? <a href="https://app.daily.dev/feed-by-ids?id=Iq2RG6GgT&amp;id=YUSdAM3D6&amp;id=d76FYhLFT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated content still influences beliefs, even after being exposed as fake</strong>: Apparently, the ‘truth crisis’ of AI has shifted from preventing initial confusion to combating persistent influence post-exposure. Research suggests people remain emotionally swayed by deepfakes even when explicitly told they are fake. So, all those ‘transparency solutions’ like Adobe’s Content Authenticity Initiative? Utterly insufficient. Because, you know, facts and critical thinking are no match for a good AI-generated lie. <a href="https://app.daily.dev/posts/savTmPDZC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Norwegian municipality’s AI billing error sends bills to children</strong>: A Norwegian municipality, in its infinite wisdom, rushed an AI implementation to correct billing errors for water and sewage fees. The result? Approximately 1,000 incorrect bills, some sent to children. The municipality admitted they ‘should have manually reviewed AI output.’ Five staff members were then required to handle the complaints. So much for saving time; it seems AI is great at generating more work, especially when deployed without a shred of common sense. <a href="https://app.daily.dev/posts/FlMxMsjGN" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
