<h2>Must know</h2>
<ul>
<li><strong>xAI faces co-founder exodus amidst deepfake controversy</strong>: xAI, Elon Musk’s AI venture, is experiencing significant internal turmoil with nine engineers, including two co-founders, publicly departing in the past week. This exodus means over half of xAI’s founding team has now left, citing desires for more autonomy and smaller teams to build frontier AI technology. These departures coincide with regulatory scrutiny over Grok’s generation of nonconsensual explicit deepfakes and the company’s upcoming IPO, raising questions about xAI’s stability and governance. <a href="https://app.daily.dev/posts/Am4GYAifG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>North Korean hackers target crypto with new macOS backdoors</strong>: North Korean threat actor UNC1069 is targeting cryptocurrency organizations with a sophisticated social engineering campaign. They’re using tactics like compromised Telegram accounts and fake Zoom meetings with AI-generated video to trick victims into executing malicious terminal commands. This deploys a multi-stage malware stack, including previously undocumented macOS backdoors like WAVESHAPER, HYPERCALL, and DEEPBREATH, designed to bypass macOS security and steal sensitive data and credentials. <a href="https://app.daily.dev/posts/WHtUJSDcB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PyTorch foundation expands to host vLLM and Ray</strong>: The PyTorch Foundation is expanding its scope beyond just PyTorch, now hosting vLLM, DeepSpeed, and Ray, aiming to become a multi-project ecosystem for critical AI infrastructure. Mark Collier, founder of OpenStack/OpenInfra, has been appointed as the new Executive Director, tasked with scaling operations and business execution. This move signifies a push for open collaboration and preventing single-entity dominance in foundational AI technologies, as reported by the PyTorch blog. <a href="https://app.daily.dev/feed-by-ids?id=2JQZDZ3HD&amp;id=CAIsBRQIQ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Microsoft .NET 11 preview 1 released</strong>: Microsoft released .NET 11 Preview 1, featuring JIT compiler improvements, CoreCLR support for WebAssembly, Zstandard compression, and C# 15 updates like collection expression arguments. <a href="https://app.daily.dev/posts/4y7L7j8R0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Erlang/OTP 29.0-rc1 brings native records and quantum crypto</strong>: Erlang/OTP 29.0-rc1 introduces native records, multi-valued comprehensions, and quantum-resistant key exchange for SSH, while removing 32-bit Windows builds. <a href="https://app.daily.dev/posts/fTCktIrzF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ESLint 10.0.0 drops .eslintrc support</strong>: ESLint 10.0.0 removes support for <code>.eslintrc</code> files, making flat config the sole configuration option, improving monorepo resolution and adding built-in TypeScript types. <a href="https://app.daily.dev/posts/eQ0JUyyzE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Broadcom unifies 177 projects into a monorepo</strong>: Broadcom successfully unified 177+ software projects into a single monorepo using Nx, achieving a 70% cache hit rate and reducing CI times to 29 minutes. <a href="https://app.daily.dev/posts/0esOYLd2k" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Garn build system boasts major speedups</strong>: Garn, an alternative Nix frontend, announced significant evaluation speed improvements using TypeScript and trace-based caching, reducing evaluation times from 800ms to 16ms in some scenarios. <a href="https://app.daily.dev/posts/edQlVXp7C" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Integrate secures $17M for defense project platform</strong>: Integrate, a Seattle startup, secured $17M in Series A funding to expand its secure collaboration platform for defense contractors and government agencies, having previously won a $25M US Space Force contract. <a href="https://app.daily.dev/posts/kwQ7z4a1U" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitGuardian raises $50M for AI agent security</strong>: GitGuardian secured $50 million in Series C funding to expand its secrets and Non-Human Identity (NHI) security platform, focusing on AI agent security and enterprise-scale NHI governance. <a href="https://app.daily.dev/posts/mmQEWnSC4" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Windsurf declares speed trumps quality in AI models</strong>: Windsurf’s latest AI model leaderboard, based on an in-IDE evaluation, claims “The People Want Speed” because faster models with “good-enough results” won more votes. Apparently, “inefficient thinking tokens” are penalized, which sounds suspiciously like a justification for their own model’s performance. So, speed over quality, because who needs deep thought when you can just get <em>something</em> quickly? <a href="https://app.daily.dev/posts/SQAix1lCw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Generative AI: the new burnout accelerator</strong>: A Harvard Business Review study found that generative AI leads to tech workers putting in longer hours, taking on more tasks, and reducing breaks, all voluntarily. Initially, it boosts productivity, but the long-term effect is workload creep, increased cognitive fatigue, and burnout. So, AI makes us work harder, not smarter, and we’re apparently happy to do it until we collapse. Progress! <a href="https://app.daily.dev/posts/mEvZkBSFL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI scrapers kill 14-year-old self-hosted git server</strong>: A developer had to shut down their self-hosted git server after 14 years because AI scrapers generated millions of 404s, overwhelming logrotate and causing a disk space outage. The developer’s blog post, titled “Thank you, AI¹,” sarcastically highlights how AI’s insatiable appetite for data can effectively DDoS small, personal infrastructure. So much for AI being a good netizen. <a href="https://app.daily.dev/posts/l8YCDT7qf" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
