<h2>Must know</h2>
<ul>
<li><strong>AI industry PACs duke it out in New York</strong>: Two rival pro-AI Super PACs are clashing in a New York congressional race, highlighting a deep split within the AI industry over regulation. One PAC, backed by Andreessen Horowitz and OpenAI’s Greg Brockman, spent $1.1 million attacking a candidate who sponsored a bill requiring AI developers to disclose safety protocols. The opposing PAC, funded by Anthropic, is supporting that candidate, showcasing a battle between minimal regulation and transparency/safety standards. <a href="https://app.daily.dev/posts/cobwYTIA0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Hugging Face and llama.cpp team up for local AI</strong>: The creators of llama.cpp, Georgi Gerganov and his team, are joining Hugging Face to ensure the long-term sustainability and growth of local AI inference. The collaboration aims to integrate llama.cpp with the Transformers library and improve packaging for easier local model deployment, all while keeping llama.cpp 100% open-source. This move is geared towards making open-source AI inference accessible on consumer hardware globally, positioning local inference as a competitive alternative to cloud solutions. <a href="https://app.daily.dev/posts/ONSshEfvD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google engineers indicted for trade secret theft</strong>: Three individuals, including two former Google engineers, have been federally indicted for allegedly stealing trade secrets related to processor security, cryptography, and Snapdragon SoC hardware architecture. Prosecutors claim the data was transferred to Iran, with the defendants using methods like photographing screens and deleting communications to evade detection. Google discovered the theft through routine security monitoring and alerted law enforcement. <a href="https://app.daily.dev/posts/hfUKRpTfu" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Xbox gets a new CEO, promises no AI-generated content</strong>: Microsoft Gaming CEO Phil Spencer and Xbox President Sarah Bond are out, with Asha Sharma, formerly of Microsoft CoreAI, taking over as the new Xbox boss, explicitly rejecting AI-generated content. <a href="https://app.daily.dev/posts/rUamL6WlB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic updates Claude Code for desktop and CLI</strong>: Anthropic released major updates to Claude Code, adding an integrated app preview, a ‘Review code’ button for inline comments and fixes, GitHub PR monitoring, and cross-device session continuity for desktop, plus git worktree support for the CLI. <a href="https://app.daily.dev/feed-by-ids?id=1pjLIO04t&amp;id=X4UvQOpsB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Red Hat launches enterprise Podman Desktop</strong>: Red Hat has launched an enterprise-supported build of Podman Desktop in technical preview, offering Docker-compatible container workflows with vendor-backed SLAs, security fixes, and tight integration with OpenShift and RHEL. <a href="https://app.daily.dev/posts/jbEPljaPx" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>EFF updates open-source policy for AI-generated code</strong>: The EFF now permits AI-generated code in open-source projects but requires human-authored comments, contributor understanding of the code, disclosure of LLM use, and accountability for design decisions. <a href="https://app.daily.dev/posts/W3yzAhjLd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Deno security vulnerabilities found by researchers</strong>: New research from NDSS 2025 identified unaddressed and partially mitigated security vulnerabilities in Deno, including ReDoS, prototype pollution, and weaknesses in its permission system, leading to Deno issuing advisories and redesigning its import mechanism. <a href="https://app.daily.dev/posts/yQbHBspCT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Dynatrace revamps mobile RUM experience</strong>: Dynatrace has overhauled its mobile Real User Monitoring (RUM) to unify performance and stability workflows, adding view-level analysis for various mobile platforms, full ANR/crash capture, and automatic source map uploading. <a href="https://app.daily.dev/feed-by-ids?id=0js0KhmpH&amp;id=AET0VSBfN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Java’s 2026 roadmap includes Valhalla, Babylon, and Loom</strong>: The Java development roadmap for 2026 includes Project Valhalla previewing value types, Project Babylon incubating code reflection, and Project Loom finalizing structured concurrency, alongside AOT compilation and new JEPs for constant patterns. <a href="https://app.daily.dev/posts/sSQwkMVM8" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>OpenAI’s ‘Frontier’ platform for enterprise AI agents</strong>: OpenAI just dropped “Frontier,” an enterprise platform for building and managing AI agents at scale. Because what every enterprise needs is more vendor lock-in and a fresh set of “Forward Deployed Engineers” to tell them how to use their own data. Apparently, shared business context and open standards are the new buzzwords for “we’ll integrate with your existing systems, eventually.” <a href="https://app.daily.dev/posts/NGz762AB9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Elon Musk delays Grok release over Baldur’s Gate performance</strong>: Elon Musk reportedly delayed a Grok model release because he wasn’t happy with its answers about Baldur’s Gate, pulling high-level engineers to fix this critical issue. TechCrunch’s “BaldurBench” (yes, really) found Grok performs comparably to rivals, proving that sometimes, even billionaires have priorities that make you scratch your head. <a href="https://app.daily.dev/posts/vj4IIdP4D" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents need human guidance, apparently</strong>: New research with something called SkillsBench (because we need more benchmarks) found that AI agents perform significantly better with “curated, human-provided skills.” Shocking, I know. Turns out, these autonomous agents aren’t so autonomous when it comes to actually being useful without us holding their digital hands. Who would’ve thought? <a href="https://app.daily.dev/posts/c3BmZNVYL" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
