<h2>Must know</h2>
<ul>
<li><strong>Apple taps Google Gemini for Siri and AI features</strong>: Apple is reportedly partnering with Google in a multi-year, non-exclusive deal to integrate Google’s Gemini AI models and cloud infrastructure into Siri and other AI features. This move, which could cost Apple around $1 billion annually according to multiple sources, comes after Apple evaluated models from OpenAI and Anthropic. A revamped, more personalized Siri is expected to launch later in 2026, with Apple emphasizing that its privacy standards will be maintained despite leveraging Google’s external services. <a href="https://app.daily.dev/posts/joVK7B34Z" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic launches agentic AI for file management</strong>: Anthropic has released Cowork, a new feature within the Claude Desktop app for macOS, available as a research preview for Claude Max subscribers. This tool allows Claude to autonomously read, edit, and create files within a specified folder via a chat interface, handling tasks like organizing receipts, analyzing data, and drafting reports. While designed for non-technical users, Anthropic acknowledges risks such as prompt injection and accidental file deletion, emphasizing user control over folder permissions. <a href="https://app.daily.dev/posts/TLkxIySjV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Hyperscale AI data centers demand next-gen nuclear power</strong>: The rapid expansion of hyperscale AI data centers, housing hundreds of thousands of GPUs, is creating an unprecedented demand for electricity, with each facility requiring over a gigawatt. This surge is driving significant interest in advanced nuclear options like small modular reactors (SMRs) that promise safer, faster, and cheaper deployment. Major tech companies like OpenAI, Google, Microsoft, and Meta are investing hundreds of billions into this infrastructure, highlighting the critical link between AI growth and energy innovation. <a href="https://app.daily.dev/feed-by-ids?id=bf0xB32y7&amp;id=qk040xtMi" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Anthropic restricts third-party access to Claude, OpenAI counters</strong>: Anthropic has limited OAuth access for third-party coding agents like OpenCode to its Claude Pro/Max subscriptions, forcing them to use higher-cost API access, while OpenAI publicly announced support for using ChatGPT subscriptions in OpenCode and similar open-source tools. <a href="https://app.daily.dev/feed-by-ids?id=8dtXe7P09&amp;id=eRi1VQ5SP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Laravel 2025 survey shows mixed adoption, new AI docs released</strong>: A 2025 Laravel developer survey revealed mixed adoption for new features like Starter kits and Laravel Cloud, with pricing and complexity being common concerns, while official AI Assisted Development documentation was released, positioning Laravel Boost as an AI-friendly toolset. <a href="https://app.daily.dev/feed-by-ids?id=12lyQYsSm&amp;id=Y3JXeC4bN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 6.19-rc5 brings NVIDIA driver and RISC-V security fixes</strong>: The latest Linux kernel release candidate, 6.19-rc5, includes a fix for newer NVIDIA GPUs in the Nouveau open-source driver, addresses a RISC-V security vulnerability, and resolves a Rust Binder bug, with two-thirds of the patch focused on driver updates. <a href="https://app.daily.dev/posts/MbuLWzK6Y" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Postman acquires Fern for enhanced API documentation and SDK automation</strong>: Postman has acquired Fern, a startup specializing in automating API documentation and generating client SDKs in nine programming languages, aiming to reduce developer friction and support costs by providing a seamless, end-to-end API lifecycle platform. <a href="https://app.daily.dev/posts/9UqlZV9z7" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes v1.35 introduces workload-aware scheduling</strong>: Kubernetes v1.35 is out, bringing workload-aware scheduling with gang scheduling to prevent deadlocks and opportunistic batching for faster pod deployment, alongside improvements in service account token handling and mutable PersistentVolume node affinity. <a href="https://app.daily.dev/feed-by-ids?id=R6fTuHjeX&amp;id=zTULpLxtX" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kotlin Multiplatform and Compose Multiplatform gain traction</strong>: Kotlin Multiplatform (KMP) and Compose Multiplatform (CMP) are enabling developers to share business logic and UI code across Android, iOS, web, and desktop, offering native performance and reducing code duplication, as demonstrated by its use in McDonald’s apps. <a href="https://app.daily.dev/posts/yGW5twYAa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Budgie desktop moves to Wayland and Qt6, IceWM sticks to X11</strong>: The Budgie 10.10 desktop environment has transitioned to native Wayland support and will re-architect Budgie 11 with the Qt6 toolkit, while the IceWM 4.0 window manager continues active development, remaining committed to the X11 display server protocol. <a href="https://app.daily.dev/posts/peCpqhbhc" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>India wants smartphone source code, government says ‘no we don’t’</strong>: India’s proposed smartphone security rules include demands for source code access, prior notification of software updates, and 12-month device log retention, which Apple, Samsung, and Google are, predictably, resisting. The IT ministry publicly denies seeking source code, calling Reuters’ reporting inaccurate, despite government documents reportedly showing explicit requirements. It’s a classic ‘trust us, not your lying eyes’ scenario, where ‘national security’ often translates to ‘we want to snoop.’ <a href="https://app.daily.dev/posts/C0BEamxdR" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI code generation: cheap to make, expensive to trust</strong>: LLM tools like Claude Code make generating code fast and accessible, even for non-developers, but apparently, the real value is still in system design and managing edge cases. Salesforce, after laying off 4,000 experienced staff to replace them with AI, reversed course three months later due to reliability issues. So, while AI can churn out code, it seems human expertise is still required to ensure it doesn’t all fall apart, proving that ‘code is cheap’ doesn’t mean ‘software is easy.’ <a href="https://app.daily.dev/feed-by-ids?id=BCKlippws&amp;id=aKgPqgLac" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft’s Copilot control: an admin’s limited fantasy</strong>: Microsoft has retired the Microsoft Deployment Toolkit (MDT) without a transition period, pushing users towards more complex, cloud-dependent solutions. Meanwhile, they’ve introduced a ‘remove Copilot’ policy for managed devices, but it’s so restrictive that admins can only remove it if it’s not user-installed and hasn’t been run in 28 days. And even then, users can just reinstall it. It’s almost as if Microsoft wants to give the illusion of control without actually giving any. <a href="https://app.daily.dev/feed-by-ids?id=0sSu27x90&amp;id=8WMJpirC0" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
