<h2>Must know</h2>
<ul>
<li><strong>Google and Apple team up on AI, Anthropic tries to lock you in</strong>: Apple is ditching OpenAI for Google’s Gemini to power a revamped Siri and other AI features across its two billion devices, paying Google around $1 billion annually. This non-exclusive partnership, reported by TechCrunch, expands an already significant relationship between the tech giants. Meanwhile, Anthropic launched Cowork, an agentic AI tool for non-coders, but then restricted third-party CLI tools like OpenCode from accessing Claude Pro/Max subscriptions, forcing developers to use their more expensive pay-per-use API. OpenAI, seizing the opportunity, immediately announced support for ChatGPT subscriptions within OpenCode and other open-source tools, as noted by The Verge. <a href="https://app.daily.dev/feed-by-ids?id=joVK7B34Z&amp;id=TLkxIySjV&amp;id=8dtXe7P09&amp;id=eRi1VQ5SP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI data centers are power hungry and nuclear is back</strong>: Hyperscale AI data centers, housing hundreds of thousands of GPUs, are consuming over a gigawatt of electricity each, comparable to small cities. Over half of this power still comes from fossil fuels, leading to significant environmental and local community impacts like rising energy costs and water shortages. In response, next-generation nuclear reactors, including small modular reactors (SMRs) and advanced fuel types, are gaining traction as a safer, more efficient way to meet this escalating energy demand, as highlighted by MIT Technology Review. <a href="https://app.daily.dev/feed-by-ids?id=bf0xB32y7&amp;id=qk040xtMi&amp;id=dJDy4tUxK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI is changing software engineering, for better or worse</strong>: GenAI tools are reshaping software engineering roles, shifting focus from coding to supervising AI-driven systems. In ‘slow flow’ organizations with weak feedback loops, these tools can accelerate defect accumulation, while ‘fast flow’ organizations with robust architectures see boosted sustainable delivery. This means AI magnifies existing organizational strengths or weaknesses, rather than solving them outright, according to reports from industry analysts. The emerging ‘Agent Experience’ (AX) paradigm prioritizes system architecture over just model improvement, with some companies even hiring junior developers again as AI raises productivity ceilings. <a href="https://app.daily.dev/feed-by-ids?id=MU6fGtVLh&amp;id=hHTJdKyX8" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>New Laravel SDK brings agentic AI to PHP</strong>: The Neuron AI Laravel SDK now allows PHP developers to create autonomous AI agents within the Laravel framework, supporting multiple AI providers like Anthropic, OpenAI, and Gemini without needing to switch to Python. <a href="https://app.daily.dev/feed-by-ids?id=4bt4Q0nem&amp;id=QVuzuBqAB&amp;id=pfLq9n48m" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google and major retailers standardize AI shopping</strong>: Google, Shopify, Etsy, Target, Walmart, and Wayfair have launched the Universal Commerce Protocol (UCP), an open standard to enable AI agents to discover products and facilitate purchases across various online retail platforms. <a href="https://app.daily.dev/feed-by-ids?id=xOjrayt7r&amp;id=2S31ZjKRT&amp;id=PLHkMQb0u&amp;id=Ywq5VNAy0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux desktop environments diverge on display servers</strong>: IceWM 4.0 continues to develop on X11 with new features, while Budgie 10.10 fully transitions to Wayland-native operation, and an experimental X11 server called Phoenix, written in Zig, aims to support modern applications with a simplified X11 protocol. <a href="https://app.daily.dev/posts/peCpqhbhc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Oracle’s 2026 Java roadmap focuses on performance and concurrency</strong>: Oracle’s Java team announced 2026 priorities for OpenJDK, including Project Valhalla for value types, Project Leyden for AOT compilation, and Project Loom for structured concurrency, all aimed at boosting Java’s speed and scalability. <a href="https://app.daily.dev/posts/gmrJODR16" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub deprecates user-to-organization account conversion</strong>: GitHub is ending the direct conversion of personal user accounts to organizations, instead introducing a ‘Move work’ flow to selectively migrate repositories to existing or new organizations while retaining personal accounts. <a href="https://app.daily.dev/posts/ush3aUToC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New Linux tool enhances battery life and CPU control</strong>: Auto-CPUFreq 3.0 for Linux introduces CPU turbo mode override via CLI and GUI, improved battery device configuration, and specific support for ASUS laptop charge thresholds, aiming to optimize battery life and performance. <a href="https://app.daily.dev/posts/umWMqgJxW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Supply chain attacks target automation platforms</strong>: Malicious npm packages impersonated n8n workflow automation integrations to steal enterprise OAuth tokens and API keys, highlighting a new vector for supply chain attacks beyond traditional build systems. <a href="https://app.daily.dev/feed-by-ids?id=E0HoVEZsQ&amp;id=UnvBaUcPK" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI health summaries: Google’s ‘dangerous’ misinformation</strong>: Google, in its infinite wisdom, decided to roll out AI Overviews for health summaries, only to have The Guardian expose them for providing ‘dangerous and inaccurate medical information’ about liver test results. Google’s response? Disable <em>some</em> problematic searches, while many others with similar risks remain. Apparently, their ‘clinician review’ process is more of a suggestion than an actual safeguard. <a href="https://app.daily.dev/posts/fofUhg4zp" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI ‘vibecoding’ security: more frameworks, less common sense</strong>: Palo Alto Networks introduced the SHIELD framework to secure AI-powered ‘vibecoding’ tools, recommending separating AI agents from production and mandatory human code reviews. Because, you know, we needed a new framework to tell us that letting an AI run wild in production without oversight is a bad idea. It’s almost like they’re discovering basic security principles all over again, but with more AI buzzwords. <a href="https://app.daily.dev/feed-by-ids?id=C2L9HfyK4&amp;id=wA63z9DYQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents: the next big security headache you didn’t ask for</strong>: Apparently, AI agents are now so autonomous they’re causing ‘unintended side effects’ and ‘exploiting intent’ with things like GeminiJack and CometJacking. Traditional security tools are useless, of course, so we need ‘semantic inspection technologies’ to analyze their ‘intent and context.’ Because nothing says progress like building systems so complex we need AI to figure out what our other AI is trying to do. <a href="https://app.daily.dev/feed-by-ids?id=CuSSpmwf6&amp;id=EDYtMcTfL&amp;id=JuPUyAfNx" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
