<h2>Must know</h2>
<ul>
<li><strong>AI coding agents are here, but at what cost?</strong>: AI coding agents are rapidly changing software development, with 85% of developers now using them and 41% of code being AI-generated. While this democratizes code creation, it introduces significant security risks like prompt injection and credential exposure, as highlighted by Security Boulevard. Organizations need robust governance and zero-trust architectures to manage these chat-based coding workflows. Meanwhile, Anthropic’s Claude Code is evolving into a terminal-based agent, but their ban on third-party API access has sparked controversy, leading to open-source alternatives like OpenCode. This shift means developers are moving from traditional IDEs to conversational interfaces, making operational excellence and Site Reliability Engineering (SRE) more critical than ever, as coding itself becomes commoditized by AI. <a href="https://app.daily.dev/feed-by-ids?id=3wv1aKzIr&amp;id=9XKw67enr&amp;id=otuC5YedQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Europe pushes for digital sovereignty, reducing reliance on US cloud giants</strong>: Europe is making significant moves to reduce its heavy dependence on US cloud providers like AWS, Azure, and Google Cloud, which currently control about 70% of the region’s market. Recent outages have underscored this vulnerability, prompting initiatives like Schleswig-Holstein’s shift to open-source alternatives and joint development of sovereign digital platforms by France, Germany, the Netherlands, and Italy. The European Commission is even introducing ‘EU Inc’ to streamline business across member states and plans a Cloud and AI Development Act to mandate data localization and stricter security for critical infrastructure, aiming to safeguard citizen data and ensure service continuity. <a href="https://app.daily.dev/feed-by-ids?id=BxeJ5ByjW&amp;id=sCcT14y3N" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Gmail’s spam filter goes rogue, causing widespread chaos</strong>: Gmail users experienced a widespread spam filtering malfunction, with legitimate emails landing in spam and spam flooding primary inboxes. Google acknowledged the issue, which affected categorization and caused significant disruption, though they’ve since stated it’s resolved. While Google didn’t disclose the technical root cause, community discussions on Hacker News speculated about recent algorithm changes. This incident highlights the critical reliance on email infrastructure and the chaos that ensues when core services misfire. <a href="https://app.daily.dev/feed-by-ids?id=I3okAWLUY&amp;id=O7yWprwoz&amp;id=cmKovtCRR" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Linux 7.0 gets Qualcomm and AMD graphics support</strong>: The upcoming Linux 7.0 kernel is set to include display support for Qualcomm Snapdragon 8 Elite Gen 5 and Adreno 225 GPUs, alongside a batch of graphics driver fixes for AMD’s RDNA4 and other GPU generations. <a href="https://app.daily.dev/feed-by-ids?id=LVxtNaVxq&amp;id=cxhDtp9rh" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PHPStan 2.1.34 delivers major performance boost</strong>: PHPStan, a static analysis tool for PHP, has released version 2.1.34, boasting 25-40% faster analysis times on real-world projects and up to a 50% reduction in runtime, significantly shortening CI pipelines. <a href="https://app.daily.dev/posts/hEVb27hAV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google releases Go package for structured LLM interactions</strong>: Google has launched jsonschema-go, a new Go package for JSON Schema, designed to define structured interactions with large language models, offering schema creation, validation, and inference from Go types. <a href="https://app.daily.dev/posts/QUtLwKAlO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Wayfair uses PostgreSQL as a dead letter queue</strong>: Wayfair has adopted PostgreSQL as a Dead Letter Queue (DLQ) for its event-driven systems, redirecting failed Kafka events to a PostgreSQL table for improved visibility, queryability, and operational simplicity in error handling. <a href="https://app.daily.dev/posts/eKm9n1imv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>FOSDEM 2026 to focus on AI, funding, and geopolitics in open source</strong>: FOSDEM 2026, expecting over 10,000 attendees, will heavily feature discussions on AI’s impact on open source, funding sustainability, and Europe’s digital sovereignty, alongside technical talks on graphics development and the Fediverse. <a href="https://app.daily.dev/feed-by-ids?id=TL2tOfQ6b&amp;id=YKcVo6svk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New macOS library for SwiftUI tab bar and split pane functionality</strong>: Bonsplit, a new native SwiftUI library for macOS, offers a customizable tab bar with split pane operations, 120fps animations, drag-and-drop reordering, and programmatic control over tab and pane states. <a href="https://app.daily.dev/posts/EwZlNqIYw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Turbopuffer ANN v3 achieves low latency on massive vector datasets</strong>: Turbopuffer’s ANN v3 system now delivers 200ms p99 query latency for 100 billion vectors at 1,000 queries per second, utilizing hierarchical clustering and binary quantization for efficient storage and recall. <a href="https://app.daily.dev/posts/EHAeemnne" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI will make us all architects, not coders. Sure, Jan.</strong>: Ryan Dahl (Node.js creator) and Linus Torvalds are apparently convinced that AI means the ‘era of humans writing code is over,’ and we’ll all just be system architects. Because, you know, AI is just going to magically handle all the messy details of actual implementation, debugging, and integrating with legacy systems. It’s a nice thought, but I’m pretty sure someone still has to get their hands dirty, and it won’t be the AI complaining about a missing semicolon. <a href="https://app.daily.dev/posts/lMnVwtEMR" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Quantum cloning breakthrough: Because the no-cloning theorem was just a suggestion</strong>: A new quantum protocol claims to enable ‘encrypted cloning’ of quantum states, supposedly circumventing the no-cloning theorem. This ‘breakthrough’ promises quantum data storage and secure communications, but conveniently omits any technical details, independent verification, or even the names of the researchers or institutions involved. So, we’re just supposed to take their word for it that a foundational principle of quantum mechanics has been casually sidestepped. Sounds legit. <a href="https://app.daily.dev/posts/z7OYuDS7t" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI’s impact on education: The ‘locked vs. open door’ dilemma</strong>: Apparently, LLMs in education present a profound ‘locked versus open door’ dilemma, offering both accessibility and the risk of undermining genuine learning. This deep insight, presented without any data, case studies, or specific examples, suggests that students might use AI to cheat instead of, you know, actually learning. Who knew? It’s almost as if technology can be used for good or ill, and we need to figure out how to teach people to use it responsibly. Groundbreaking. <a href="https://app.daily.dev/posts/2lqHEVp1m" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
