<h2>Must know</h2>
<ul>
<li><strong>AI agent security is a dumpster fire</strong>: AI agents are multiplying faster than rabbits, with one social network for AI agents, Moltbook, revealing an 88:1 agent-to-human ratio and a misconfigured database exposing 1.5 million API keys. This highlights a critical issue: AI-coded applications prioritize speed over secure defaults. Meanwhile, security researchers found critical vulnerabilities in OpenClaw, an open-source AI agent platform, including data exfiltration and exposed credentials, with thousands of instances discoverable online, according to multiple reports. <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>Microsoft’s AI push is causing more problems than solutions</strong>: Microsoft’s aggressive AI integration in Windows 11 is facing significant user backlash, leading them to review and potentially remove Copilot integrations from apps like Notepad and Paint. This comes as Microsoft’s emergency patch frequency for Windows has significantly increased, coinciding with claims that over 30% of its code is now AI-generated, raising serious questions about quality control, as reported by multiple sources. <a href="https://app.daily.dev/feed-by-ids?id=H1aWwdchu&amp;id=0jt1kGbdr&amp;id=8b12pEovl&amp;id=mR53bImT0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Raspberry Pi prices are up thanks to AI</strong>: Raspberry Pi announced price increases of $10 to $60 for models with 2GB or more memory, including the Pi 4 and Pi 5. The culprit? Increased LPDDR4 memory costs, driven by AI infrastructure demand for manufacturing capacity. So, while you’re building your self-hosted open-source AI chatbot, remember you’re contributing to the problem. <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 workflows</strong>: OpenAI released a new specification to standardize agentic AI workflows, aiming to eliminate API fragmentation and allow developers to switch between models without rewriting integration code. <a href="https://app.daily.dev/posts/yIvolp9mz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New open-source AI models and frameworks emerge</strong>: Moonshot AI released Kimi K2.5, a 1 trillion parameter multimodal AI model, while the Gradio team launched Daggr, an open-source Python library for building and debugging multi-step AI workflows. <a href="https://app.daily.dev/feed-by-ids?id=45kpQHTQG&amp;id=CoNCaDnyA&amp;id=L2SMBoUpw&amp;id=qG0ZTfTpe&amp;id=qKtGa6cI1&amp;id=w5uxAjhHS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub addresses low-quality and AI-generated contributions</strong>: GitHub is implementing repository-level controls and planning AI-powered triage tools to combat the surge of low-quality pull requests, including AI-generated code, overwhelming open-source maintainers. <a href="https://app.daily.dev/posts/jwftmMKjS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Firefox 148 introduces granular AI controls</strong>: Firefox 148 will include a new AI Controls panel, allowing users to disable AI features at a granular level or block them entirely, following user backlash against unwanted AI integrations. <a href="https://app.daily.dev/posts/K3PNUlsQj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft NTLM authentication to be disabled by default</strong>: Microsoft will disable the outdated NTLM authentication protocol by default in Windows 11 and Windows Server by 2026, a move aimed at improving security but potentially impacting legacy enterprise systems. <a href="https://app.daily.dev/posts/ufefrEjrA" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Git 2.53.0 released with geometric repacking and improved fast-import</strong>: Git 2.53.0 introduces geometric repacking 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>Linux From Scratch drops SysVinit support for systemd-only future</strong>: Linux From Scratch (LFS) will become systemd-only starting with version 13.0, citing overwhelming volunteer workload and increasing upstream dependencies on systemd from major desktop environments. <a href="https://app.daily.dev/posts/ohWaYbFxO" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-generated content persists despite being exposed as fake</strong>: Apparently, people are still emotionally swayed by deepfakes even after being told they’re fake. So, all those ‘transparency solutions’ like Adobe’s Content Authenticity Initiative? Utterly useless. It’s not about preventing initial confusion, it’s about combating persistent influence. Who knew humans were so irrational? <a href="https://app.daily.dev/posts/savTmPDZC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Manis AI platform promises mobile apps in 8 minutes</strong>: A new no-code platform, Manis, claims to build and deploy mobile apps in about 8 minutes using AI orchestration. They even demonstrated a ‘Bill Sniper’ app from a single prompt. Because what the world really needs is more hastily built, AI-orchestrated mobile apps with one-click deployment to Google Play. What could possibly go wrong? <a href="https://app.daily.dev/posts/mN5yBPuXJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cloudflare’s ‘production-grade’ matrix homeserver was just AI-generated fluff</strong>: Cloudflare proudly announced a ‘production-grade’ Matrix homeserver built on Workers, only for the Matrix co-founder to point out it lacked critical features and contained unreviewed AI-generated code. Cloudflare later added a disclaimer, reclassifying it as a ‘proof of concept.’ It’s almost like AI-generated code needs human oversight, or maybe, just maybe, it’s not ‘production-grade’ if it’s full of TODOs and missing basic functionality. <a href="https://app.daily.dev/feed-by-ids?id=46vOfTKm4&amp;id=IgyIzqLcs&amp;id=Iq2RG6GgT&amp;id=bREDMGnKC&amp;id=d6wB2Rm36" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
