<h2>Must know</h2>
<ul>
<li><strong>Meta secures massive nuclear power deals for AI data centers</strong>: Meta has inked agreements with Vistra, Oklo, and TerraPower to secure over six gigawatts of nuclear power for its data centers, driven by escalating AI workloads. Vistra will immediately supply 2.1 GW from existing Ohio facilities, with an additional 433 MW by the early 2030s. Oklo and TerraPower, both advanced nuclear startups, are slated to deliver 1.2 GW and 2.8 GW respectively, targeting operational dates between 2030-2032. This move signals a major shift in the tech sector towards long-term nuclear energy procurement, though the cost-effectiveness of small modular reactors (SMRs) remains unproven, as reported by TechCrunch. <a href="https://app.daily.dev/posts/1sXfoaLc2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Snowflake acquires Observe for AIops and observability</strong>: Snowflake is acquiring Observe, an AI-based Site Reliability Engineering (SRE) platform, to enhance its observability offerings. The goal is to unify monitoring and management of AI systems, data pipelines, and infrastructure health within a single platform. This acquisition aims to address challenges in diagnosing performance and data anomalies in AI-driven applications, leveraging Snowflake’s Data Cloud for cost-effective telemetry data management, according to Snowflake’s official announcement. <a href="https://app.daily.dev/posts/XlAd95QPv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>North Korean hackers exploit QR codes for credential theft</strong>: The FBI attributes recent malicious QR code spear-phishing campaigns to North Korean state-sponsored hackers, specifically Kimsuky/APT43. These attacks target U.S. think tanks, academic institutions, and government entities, using fake login pages for Microsoft 365, Okta, and VPNs to harvest credentials and bypass multi-factor authentication via session token theft. The use of ‘quishing’ exploits the fact that mobile devices often operate outside traditional enterprise security tools, as detailed in FBI advisories. <a href="https://app.daily.dev/posts/XmPhFlPf5" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Anthropic expands enterprise AI market share with Allianz partnership</strong>: Anthropic has partnered with Allianz to integrate AI tools, including Claude Code and custom AI agents, across the insurer’s workforce, claiming a 40% enterprise AI market share based on a Menlo Ventures survey. <a href="https://app.daily.dev/feed-by-ids?id=D8jNyiFSs&amp;id=DLaOdwegi" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Non-human identity security becomes critical for AI agents</strong>: The rapid adoption of AI agents is increasing the volume and complexity of non-human identities (NHIs), exposing unmanaged credentials and broad permissions, making NHI security a critical concern for finance, healthcare, and DevOps. <a href="https://app.daily.dev/feed-by-ids?id=0XmbFmo97&amp;id=43f4JcvZ9&amp;id=8827svTqM&amp;id=Jc3Gb3RMH&amp;id=kl0LpaKDU&amp;id=lqqfnNpcw&amp;id=xc8AsvvNa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google overhauls Gmail with new AI features</strong>: Google is transforming Gmail into a “proactive inbox assistant” with new AI features like AI Overviews for email summaries, natural language inbox queries, and enhanced writing assistance, all powered by Gemini 3. <a href="https://app.daily.dev/posts/8lOS5k1vH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>China investigates Meta’s acquisition of Manus AI</strong>: Chinese authorities are investigating Meta’s acquisition of Manus, an AI platform that relocated from China to Singapore, for potential violations of export controls and foreign investment laws, signaling tightening regulatory scrutiny on Chinese-founded tech firms. <a href="https://app.daily.dev/posts/Z4bL3wwsW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust gains traction for AI and deep learning frameworks</strong>: Nathaniel Simard, creator of the Burn deep learning framework, advocates for Rust in AI and high-performance computing, citing its type system, ownership rules, and ability to manage complex codebases for reliability and maintainability. <a href="https://app.daily.dev/feed-by-ids?id=JPj2yTBJd&amp;id=PsAbhn9Cz&amp;id=inqrenkKY" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Canonical releases ARM64 Steam Snap with x86 emulation</strong>: Canonical has released a Steam Snap for Ubuntu ARM64, bundling the x86 Steam client with the FEX emulator, allowing users to run x86 Steam games on ARM64 Ubuntu systems, though it’s experimental and performance depends on high-end hardware. <a href="https://app.daily.dev/posts/Z9Oa5uBqK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Zed editor’s December update brings performance and AI improvements</strong>: Zed, the Rust-based editor, released v0.219.0 with 64 closed issues and 169 merged PRs, featuring editor enhancements, performance optimizations like direct-to-display rendering on macOS, and AI features including model favoriting and collapsible context badges. <a href="https://app.daily.dev/posts/AqqiPNIq8" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-friendly documentation: the new gold rush for SEO?</strong>: Apparently, we now need to make our documentation ‘AI-friendly’ with an <code>llms.txt</code> file, because good old plain language and semantic HTML aren’t enough for our silicon overlords. The idea is to keep documentation current so AI models don’t just hallucinate outdated info, but let’s be real, this sounds like another excuse to optimize for bots instead of actual humans. Just write clear docs, folks; the AI will figure it out, or it won’t, and then we’ll have bigger problems. <a href="https://app.daily.dev/posts/RuMhQtj6W" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code quality: more code, more problems?</strong>: SonarSource’s leaderboard shows some LLMs generate Java code with 3x more lines and significant security vulnerabilities, despite claims of efficiency. While Opus 4.5 Thinking might be decent, others are churning out verbose, insecure garbage. So, while AI can generate code faster, it seems we’re just trading developer time for more time debugging AI’s ‘creative’ interpretations of secure coding practices. Always run those SAST scanners, unless you enjoy SQL injection roulette. <a href="https://app.daily.dev/posts/E02dc9OJZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GameStop’s grand strategy: close everything, pay CEO for magic</strong>: GameStop is closing 390 stores in the US, exiting four European countries, and trying to sell off operations in France and Canada, all while net sales plummeted 27.5%. But don’t worry, CEO Ryan Cohen got a $35 million performance incentive, contingent on raising the market cap to a cool $100 billion. Because when your business is collapsing, the best strategy is to cut everything and hope for a miracle, preferably one that makes the CEO very rich. Solid plan. <a href="https://app.daily.dev/posts/t6Y3Ssj5q" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
