<h2>Must know</h2>
<ul>
<li><strong>Google’s Gemini AI faces IP theft and nation-state exploitation</strong>: Google blocked over 100,000 attempts to extract proprietary reasoning capabilities from its Gemini AI model, which it considers intellectual property theft. Simultaneously, Google’s threat report revealed nation-state actors are integrating Gemini into attack operations for social engineering, vulnerability analysis, and phishing. New malware families like HONESTCUE are even embedding Gemini’s API to generate and execute malicious code, bypassing safety filters. (Source: Google’s internal detections and threat report) <a href="https://app.daily.dev/posts/OQLlwLpSw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>TypeScript 6.0 beta is the last JavaScript-based release</strong>: TypeScript 6.0 beta is the final version built on JavaScript, as TypeScript 7.0 will transition to a Go-based compiler. This is a significant architectural shift, aiming for better performance and faster build times. The beta includes a <code>--stableTypeOrdering</code> flag to help with migration, support for <code>es2025</code> and new Temporal API types, and improvements to <code>this-less</code> function handling. (Source: TypeScript release notes) <a href="https://app.daily.dev/posts/62kNn8zlZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 7.0 integrates quantum-resistant cryptography and performance boosts</strong>: Linux 7.0 is bringing a host of performance optimizations, including improved memory management for AMD Zen architectures and significant speedups for XFS file system operations. Crucially, it’s integrating ML-DSA (Dilithium), a quantum-resistant signature algorithm, for module signing, replacing SHA-1. This move is part of a broader effort to migrate to post-quantum cryptography, addressing the threat quantum computers pose to current encryption standards. (Source: Linux kernel development community) <a href="https://app.daily.dev/feed-by-ids?id=6DJe6t3uU&amp;id=Ukr75F53X&amp;id=Vay0ocYOF&amp;id=Vh5TboDqs&amp;id=m5L4rWAXQ&amp;id=ppwBwOuN5&amp;id=uGdXnDZH4&amp;id=xVcPYGmKq" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>OpenAI launches GPT-5.3-Codex-Spark on Cerebras hardware</strong>: OpenAI released GPT-5.3-Codex-Spark, a new coding model running on Cerebras CS3 accelerators, achieving over 1,000 tokens per second, prioritizing speed over depth. <a href="https://app.daily.dev/feed-by-ids?id=2DKu3o06c&amp;id=2G5oHg0oJ&amp;id=A9s0KznyU&amp;id=CacFMCP2L&amp;id=PFJ429xnc&amp;id=TgVTzT7ZG&amp;id=dqsxZt8l5&amp;id=f8bwmVFOm&amp;id=iVwXrkdSj&amp;id=yYXkpJGbb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Go 1.26 released with Green Tea GC and <code>cgo</code> improvements</strong>: Go 1.26 is out, featuring the Green Tea garbage collector enabled by default, a new <code>new</code> syntax for initial values, and a 30% reduction in <code>cgo</code> call overhead. <a href="https://app.daily.dev/feed-by-ids?id=1ObNpwyIZ&amp;id=mj2oUwkea" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Shadcn releases visual project builder for custom setups</strong>: Shadcn launched a visual project builder (<code>npx shadcn create</code>) allowing developers to customize themes, components, and design systems for Next.js, Vite, and TanStack Start with real-time previews. <a href="https://app.daily.dev/posts/1PTkZmoEH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Orchestra: a new self-hosted PaaS for bare metal</strong>: Orchestra, a new open-source, self-hosted PaaS, automates server provisioning, cluster orchestration (Kubernetes/K3s, Docker Swarm), and application deployment on bare metal without agents. <a href="https://app.daily.dev/feed-by-ids?id=0nUdXf1b9&amp;id=C2t5MyKnp&amp;id=jNBhBoH4V&amp;id=nwGZvk6OS&amp;id=sjdf7M4Ia" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rails updates bring transaction customization and performance boosts</strong>: Rails received updates including a new extension point for customizing transactions, performance improvements for <code>filter_parameters</code>, and fixes for Marshal deserialization and composite primary keys. <a href="https://app.daily.dev/feed-by-ids?id=5hfEVLtRT&amp;id=CWjTT1lwV&amp;id=oKSotbpAp" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>DIY project digitizes childhood videos with AI and Supabase</strong>: A developer built a custom pipeline to digitize 200+ hours of childhood footage, using FireWire on Linux, AI-assisted transcription (WhisperX), ffmpeg for HLS, and Cloudflare R2/Supabase for a family streaming platform. <a href="https://app.daily.dev/posts/zquW7brd7" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PVS-Studio to release static analyzer for Go</strong>: PVS-Studio is launching a static analyzer for Go in April 2026, claiming to catch bugs that <code>go vet</code> misses, with examples found in Docker, Kubernetes, and Gitea. <a href="https://app.daily.dev/feed-by-ids?id=1ObNpwyIZ&amp;id=mj2oUwkea" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Minimax M2.5: the cheapest and best AI model this week</strong>: Minimax launched its M2.5 frontier AI model, claiming it’s over 90% cheaper than Claude Opus or GPT-5 models at $1/hour. OpenHands benchmarked it as the top open model for coding tasks. Because, you know, a new AI model is always the cheapest and best, until the next one drops next week. (Source: Minimax, OpenHands) <a href="https://app.daily.dev/posts/VQt2ETdNG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s AI-generated C compiler: more hype than substance</strong>: Anthropic’s AI-generated C compiler, built by 16 Claude Opus agents in two weeks, is being touted as a breakthrough. However, critics point out it fails basic tests and relies on existing tools. So, it’s less a revolutionary compiler and more a fancy marketing demo. But hey, it sounds good on a press release, right? (Source: mZHhw9XzP) <a href="https://app.daily.dev/feed-by-ids?id=0wFgrbFqP&amp;id=78oOrNVDz&amp;id=B3N2Icy7Y&amp;id=BKJ8NHw6d&amp;id=DcSKM9BQA&amp;id=GManH4sq3&amp;id=GnwSkM7qk&amp;id=HHMxTl4yN&amp;id=JgHE2Ywe8&amp;id=MPMZLgvp3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Gartner predicts 40% AI project failure: shocker</strong>: Gartner predicts 40% of AI projects will fail by 2027 due to “unsustainable velocity, fragmented systems, and shadow AI.” Apparently, throwing AI at every problem without a coherent strategy or proper governance isn’t a recipe for success. Who knew? It’s almost like these things require actual planning and not just buzzwords. (Source: Gartner) <a href="https://app.daily.dev/feed-by-ids?id=NUfU1KApc&amp;id=Ss5yJGdsu&amp;id=ZY9MBJa3v&amp;id=myl2tyVQW&amp;id=sK6lOaq5D" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
