<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 open-source, with Hugging Face providing resources to scale the project. This move aims to make open-source AI accessible on consumer hardware, 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>OpenAI and Anthropic push enterprise AI agent platforms</strong>: OpenAI has launched “Frontier,” an enterprise platform for building and managing AI agents at scale, integrating with existing systems and offering governance controls. This move, however, has sparked community pushback regarding OpenAI’s enterprise pivot and concerns about vendor lock-in. Meanwhile, Anthropic updated Claude Code with new desktop capabilities, including an integrated app preview, code review features, and GitHub PR monitoring, alongside adding git worktree support to its CLI for parallel agent operation. <a href="https://app.daily.dev/feed-by-ids?id=NGz762AB9&amp;id=1pjLIO04t&amp;id=X4UvQOpsB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>EFF allows AI-generated code in open source with caveats</strong>: The EFF has updated its open-source contribution policy to permit AI-generated code, provided contributors ensure human-authored documentation, understand the code, and disclose LLM use. This pragmatic approach acknowledges the impracticality of banning AI-assisted coding and the unsolvable problem of AI detection, shifting accountability to the developer to explain and test their submissions. This policy aims to mitigate concerns about low-quality AI contributions and potential hallucinations. <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>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 and tight OpenShift integration. <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 human-provided skills, showing no improvement when generating their own. <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 prison for operating a platform that sold stolen US identities to North Korean IT workers, funneling 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>: Google detected and alerted law enforcement to a federal indictment against three individuals, including two former Google engineers, for allegedly stealing trade secrets related to processor security and cryptography, transferring 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 exposed by research</strong>: Deno’s security claims against Node.js were challenged by NDSS 2025 research, which found unaddressed vulnerabilities like ReDoS 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>Undocumented accelerometer access on Apple Silicon MacBooks</strong>: A Python project by olvvier can read raw 3-axis acceleration data from undocumented MEMS accelerometers on Apple Silicon MacBooks, enabling experimental applications like heart rate estimation from chassis vibrations. <a href="https://app.daily.dev/posts/zUsdwvEbb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Erlang/OTP patch fixes critical vulnerabilities</strong>: The Erlang/OTP 27.3.4.8 patch release fixes a path traversal vulnerability in the tftp module, a crash-inducing bug, and a memory barrier issue on ARM architectures. <a href="https://app.daily.dev/posts/5THfLlOSG" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI industry PACs battle over New York congressional race</strong>: Apparently, two rival pro-AI Super PACs are duking it out in a New York congressional race, spending millions to either attack or support a candidate based on their stance on AI regulation. Because nothing says “responsible AI development” like throwing cash at politicians to avoid pesky safety protocols. It’s almost as if some in the industry prefer to operate in a regulatory vacuum, while others pretend to care about safety with their own PACs. <a href="https://app.daily.dev/posts/cobwYTIA0" 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 what the world really needs is more half-baked AI projects from local meetups. It’s a brilliant strategy to offload the actual development work to the community while OpenAI focuses on… well, whatever it is they’re doing with their enterprise platform. <a href="https://app.daily.dev/posts/kpaXVeVV9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Grok release delayed over Baldur’s Gate performance</strong>: Elon Musk reportedly delayed a Grok model release because he was unhappy with its answers about the video game Baldur’s Gate, pulling high-level engineers to fix the issue. Forget curing cancer or solving climate change; the real priority is ensuring your AI can properly discuss fantasy RPGs. TechCrunch’s “BaldurBench” then found Grok performs “comparably” to rivals, proving that a billionaire’s video game preferences are indeed a valid benchmark for AI progress. <a href="https://app.daily.dev/posts/vj4IIdP4D" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
