<h2>Must know</h2>
<ul>
<li><strong>Llama.cpp joins Hugging Face for local AI growth</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. They will retain full technical autonomy, and llama.cpp will remain 100% open-source. Hugging Face will provide resources to scale the project, with a shared vision to make open-source AI accessible on consumer hardware, aiming to make local inference 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>AI industry super PACs clash over regulation in NY race</strong>: Two rival pro-AI Super PACs are clashing in a New York congressional race, highlighting a significant split within the AI industry regarding 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 direct political battle over AI’s future. <a href="https://app.daily.dev/posts/cobwYTIA0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>EFF updates open-source policy for AI-generated code</strong>: The EFF has updated its open-source contribution policy to permit AI-generated code, acknowledging that banning it is impractical and detection is an unsolvable problem. The policy requires contributors to disclose LLM use, thoroughly understand and test any AI-generated code, and ensure all comments and documentation are human-authored. This shifts accountability upstream, requiring contributors to explain design decisions and justify choices, with enforcement relying on trust and spot checks. <a href="https://app.daily.dev/posts/W3yzAhjLd" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Claude Code Desktop gets major update</strong>: Anthropic released a major update to Claude Code Desktop, adding features like an integrated app preview, a ‘Review code’ button for inline comments and auto-fixes, and GitHub PR monitoring with optional auto-fix for CI failures. <a href="https://app.daily.dev/posts/1pjLIO04t" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI launches enterprise AI agent platform</strong>: OpenAI launched “Frontier,” an enterprise-focused platform for building, deploying, and managing AI agents at scale, integrating with existing systems and offering Forward Deployed Engineers for adoption support. <a href="https://app.daily.dev/posts/NGz762AB9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Red Hat launches enterprise Podman Desktop</strong>: Red Hat launched an enterprise-supported build of Podman Desktop in technical preview, targeting corporate developers with Docker-compatible container workflows, tight OpenShift integration, and features for fleet-wide configuration management. <a href="https://app.daily.dev/posts/jbEPljaPx" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Human guidance critical for AI agent performance</strong>: New research using the SkillsBench benchmark indicates that AI agents perform significantly better with curated, human-provided skills, showing no improvement when generating their own skills. <a href="https://app.daily.dev/posts/c3BmZNVYL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ukrainian sentenced for aiding North Korean IT worker fraud</strong>: A Ukrainian national was sentenced to five years in US federal prison for operating a platform that sold stolen US identities to North Korean IT workers, enabling them to fraudulently secure employment and funnel earnings to North Korea’s weapons program. <a href="https://app.daily.dev/posts/icCY5Agrr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Federal indictment for Google trade secret theft</strong>: Three individuals, including two former Google engineers, were federally indicted for stealing trade secrets related to processor security, cryptography, and Snapdragon SoC architecture, allegedly transferring the data to Iran. <a href="https://app.daily.dev/posts/hfUKRpTfu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Deno security vulnerabilities identified by NDSS research</strong>: Deno’s security claims were scrutinized by NDSS 2025 research, which found unaddressed vulnerabilities like ReDoS, partially mitigated issues like 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>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Grok release delayed over Baldur’s Gate performance</strong>: Apparently, Elon Musk delayed a Grok model release because he wasn’t happy with its answers about the video game Baldur’s Gate, even pulling high-level engineers to fix it. Business Insider reported on this, and TechCrunch’s “BaldurBench” found Grok now performs comparably to rivals. So, the future of AI hinges on its ability to discuss D&amp;D lore, who knew? <a href="https://app.daily.dev/posts/vj4IIdP4D" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ruby is the new AI darling, apparently</strong>: A new report suggests Ruby is a “strong contender” for AI application development, specifically for LLM integration, because modern AI is just “web engineering around HTTP API calls.” The article, from an unnamed source, claims Python’s ML ecosystem is “less relevant.” So, all those Pythonistas building actual models are apparently just wasting their time, because, you know, HTTP calls are the real innovation. <a href="https://app.daily.dev/posts/5tjEzlHaH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI promotes Codex developer meetups</strong>: OpenAI is promoting community meetups for its Codex technology, encouraging developers to “create and ship projects.” Because nothing says innovation like a good old-fashioned meetup, especially when it’s for a technology that’s been around for a while. It’s almost like they’re trying to remind us that Codex still exists, or perhaps they’re just really into local coffee shops. <a href="https://app.daily.dev/posts/kpaXVeVV9" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
