<h2>Must know</h2>
<ul>
<li><strong>Microsoft’s AI spending spree and Windows 11 woes</strong>: Microsoft’s stock took a hit after reporting a massive $37.5 billion in AI-related capital expenditure, raising questions about the ROI of their aggressive AI push. Critics, like those on Hacker News, suggest the focus on AI has led to a neglected and bloated Windows 11, with Copilot integrations feeling forced. This comes as Microsoft also faces a troubling trend of increasing emergency patches for Windows, coinciding with workforce reductions and claims of significant AI-contributed code, hinting at potential quality control issues. <a href="https://app.daily.dev/feed-by-ids?id=0PRBbgRZc&amp;id=H1aWwdchu&amp;id=0jt1kGbdr&amp;id=8b12pEovl&amp;id=mR53bImT0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent security is a dumpster fire</strong>: Security researchers are sounding the alarm on critical vulnerabilities in AI agent platforms. A misconfigured Supabase database at Moltbook, an AI social network, exposed 1.5 million API keys and private messages, revealing an 88:1 ratio of AI agents to human users with no verification. Furthermore, OpenClaw, a popular open-source AI agent project, is riddled with security flaws like data exfiltration and exposed credentials, with thousands of instances discoverable on Shodan. This highlights a dangerous trend of prioritizing development speed over secure defaults in the AI agent ecosystem. <a href="https://app.daily.dev/feed-by-ids?id=iW2dnVT4l&amp;id=nVdzT2QRW&amp;id=fuL6WfKHt&amp;id=IVsgr78CO&amp;id=SulJvcnOu&amp;id=kgAIr1pL1&amp;id=puWFcrOO1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Raspberry Pi prices jump due to AI memory demand</strong>: Raspberry Pi has increased prices for models with 2GB or more memory, with hikes ranging from $10 to $60. The company attributes this directly to increased LPDDR4 memory costs, driven by the insatiable demand from AI infrastructure. While models with 1GB RAM remain unaffected, this marks a continuing trend of supply chain pressures impacting even low-cost hardware due to the broader AI boom. <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 releases open responses specification for agentic AI</strong>: OpenAI has released the Open Responses specification to standardize agentic AI workflows and eliminate API fragmentation across LLM providers, enabling developers to switch models without rewriting integration code. <a href="https://app.daily.dev/posts/yIvolp9mz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Nvidia-OpenAI investment uncertainty causes stock drop</strong>: Nvidia’s stock dropped following reports of uncertainty regarding its planned $100 billion investment in OpenAI, amidst concerns about circular financing in AI and Nvidia’s strategic balancing act. <a href="https://app.daily.dev/posts/isSX8DfZa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MongoDB Atlas launches embedding and reranking API</strong>: MongoDB Atlas introduced a public preview of its Embedding and Reranking API, integrating Voyage AI’s search models directly into the platform to simplify building Retrieval Augmented Generation (RAG) systems. <a href="https://app.daily.dev/posts/Ow0bpZEWc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Firefox 148 introduces granular AI controls</strong>: Firefox 148 will feature an AI Controls panel, allowing users to disable AI features at a granular level or entirely, following user backlash against CPU-intensive and commercially-driven AI integrations. <a href="https://app.daily.dev/posts/K3PNUlsQj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent knowledge sharing platforms emerge</strong>: New platforms like Agent Knowledge Commons and Reposit are emerging to enable AI agents to share solutions, learn collectively, and avoid redundant work, indicating active development in the AI agent ecosystem. <a href="https://app.daily.dev/feed-by-ids?id=Jq25xqmJD&amp;id=YjbPdZkNb&amp;id=j0iRlHvmi" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Git 2.53.0 released with geometric repacking</strong>: Git 2.53.0 introduces geometric repacking support for partial clone repositories and a new ‘strip-if-invalid’ mode for git-fast-import, enhancing performance and security. <a href="https://app.daily.dev/posts/19pgRer8D" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vaadin introduces Swing modernization toolkit</strong>: Vaadin announced a Swing Modernization Toolkit, enabling Java Swing applications to run in web browsers and allowing incremental modernization into web apps without risky full rewrites. <a href="https://app.daily.dev/posts/00Qm1Wvf0" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI manipulated content persists despite exposure</strong>: Apparently, AI-generated fake content continues to sway people even after being exposed as fake. The article ‘What we’ve been getting wrong about AI’s truth crisis’ suggests transparency alone isn’t enough. So, we’re not just dealing with AI that lies, but AI that lies so well, people still believe it even when they know it’s a lie. Good luck with that ‘truth crisis,’ folks. <a href="https://app.daily.dev/posts/savTmPDZC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI coding tools: the productivity paradox</strong>: A Norwegian developer warns of an ‘AI bubble’ bursting, leading to higher prices and lower quality AI services. Meanwhile, some claim AI tools accelerate code generation but paradoxically reduce overall productivity by making complexity cheap and compounding organizational chaos. So, AI is simultaneously making us faster and slower, better and worse. Sounds like a perfectly balanced, totally not overhyped, future. <a href="https://app.daily.dev/feed-by-ids?id=9HijPHSEH&amp;id=AIQanc1ui&amp;id=NEIMwAdHs&amp;id=k9rFokMvG&amp;id=q4q4xGaBF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Manis AI platform builds mobile apps in 8 minutes</strong>: Manis, a no-code platform, claims to build and deploy mobile apps in about 8 minutes using AI orchestration. The article, titled ‘I Built &amp; Launched a Mobile App in 10 Minutes (No Code),’ frames this as a personal success story. Because, clearly, the quality, maintainability, and long-term viability of an app built in less time than it takes to brew coffee are totally irrelevant. Just click and deploy, what could possibly go wrong? <a href="https://app.daily.dev/posts/mN5yBPuXJ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
