<h2>Must know</h2>
<ul>
<li><strong>Malicious code in Aqua Trivy VS Code extension spawns rogue AI agents</strong>: Socket researchers discovered that versions 1.8.12 and 1.8.13 of the popular Aqua Trivy extension were compromised to silently launch AI coding assistants like Claude and GitHub Copilot in unattended, fully permissive modes. The injected code used natural-language prompts to trick these agents into performing system reconnaissance and exfiltrating credentials to a rogue GitHub repository. If you have these versions installed, rotate your tokens immediately and audit your AI agent logs for unauthorized activity. <a href="https://app.daily.dev/posts/DGWq9i7NL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic and OpenAI split over Pentagon AI ethics</strong>: OpenAI has secured a deal allowing the US military to use its models in classified settings, framing it as a compromise that prohibits autonomous weaponry but allows for data analysis. Conversely, Anthropic lost its Pentagon contract due to a stricter stance on military use, leading Defense Secretary Hegseth to threaten classifying the company as a supply chain risk. This divergence highlights a growing rift in how major AI labs navigate the lucrative but ethically fraught defense sector. <a href="https://app.daily.dev/feed-by-ids?id=WNUStVFJw&amp;id=YzS181cNy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Motorola breaks Pixel monopoly on GrapheneOS</strong>: In a major shift for mobile privacy, Motorola announced a partnership at MWC to officially support GrapheneOS on its upcoming flagship hardware, including the razr fold and ThinkPhone lines. This marks the first time a major OEM has committed to meeting GrapheneOS’s strict hardware security requirements, such as memory tagging and Titan-equivalent security guarantees. While current hardware isn’t compatible, Motorola plans to ship GrapheneOS-ready devices starting in 2027 to target enterprise and government sectors. <a href="https://app.daily.dev/posts/InYjzvxra" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Mercury 2 launches as the first production diffusion-based LLM</strong>: Inception Labs released Mercury 2, which uses a diffusion architecture to generate over 1,000 tokens per second, significantly outperforming transformer-based models in raw speed. <a href="https://app.daily.dev/posts/Vqh5XymdW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Official ClickHouse .NET client hits stable 1.0.0</strong>: The new driver introduces a thread-safe ClickHouseClient API, JWT auth, and native support for QBit vectors, while dropping support for legacy .NET Framework. <a href="https://app.daily.dev/posts/u8peM5Nox" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub Copilot implements subscription-based network routing</strong>: Enterprise and Business users must update their firewall allowlists to include new specific endpoints like <a href="http://api.enterprise.githubcopilot.com" target="_blank" rel="noopener nofollow">api.enterprise.githubcopilot.com</a> to maintain connectivity. <a href="https://app.daily.dev/posts/SUKA5FOXy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WiFi DensePose performs human tracking without cameras</strong>: This Rust-based open-source system uses commodity WiFi signals to estimate human poses and monitor vital signs through walls at 54,000 frames per second. <a href="https://app.daily.dev/posts/k6VPJ0C37" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Apple M4 iPad Air arrives with 12GB unified memory</strong>: The updated tablet features the M4 chip with a 3x faster Neural Engine for on-device AI and starts at $599 for the 11-inch model. <a href="https://app.daily.dev/feed-by-ids?id=JiFSBsNyC&amp;id=GaSTgZRqL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>TanStack Start benchmarks 83 percent faster than Next.js 16</strong>: In production SaaS testing, TanStack Start demonstrated significantly faster dev server startups and lower memory leak risks compared to the latest Next.js. <a href="https://app.daily.dev/posts/ed7TZMhpP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 7.1 to power off systems on fatal ACPI errors</strong>: A new kernel change will align with ACPI specs by automatically shutting down hardware upon fatal errors, though it can be disabled via boot flag. <a href="https://app.daily.dev/posts/TLZKRpI81" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Tailwind is dying because AI stopped people from reading docs</strong>: Industry newsletters are claiming Tailwind CSS is in a ‘financial death spiral’ because AI agents are answering developer questions, causing a 40% drop in documentation traffic and ad revenue. It’s a classic case of blaming the ‘AI boogeyman’ for a business model that relied on people manually scrolling through utility class lists instead of just building a sustainable product. Expect Vercel to ‘heroically’ acquire them any minute now to save the ecosystem they helped complicate. <a href="https://app.daily.dev/feed-by-ids?id=o7ayyfPeQ&amp;id=peaMJGK0d" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Atlassian claims 60 minute meetings solve AI alignment</strong>: Atlassian’s ‘Teamwork Lab’ reports a 34% increase in AI alignment after a single one-hour retrospective. If you believe a 60-minute chat can fix the fundamental hallucination and security risks of agentic AI, I have a bridge in Jira to sell you. It’s peak corporate ‘vibes-based’ metrics designed to make middle management feel like they’re steering the ship while the AI just does whatever it was going to do anyway. <a href="https://app.daily.dev/posts/s9jeiAAAG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The senior engineer talent gap is an AI problem</strong>: Microsoft execs are warning that AI coding tools are ‘narrowing the talent pyramid’ by preventing juniors from learning the basics. This is a convenient narrative that ignores a decade of companies refusing to hire juniors in the first place. Blaming Copilot for the lack of senior engineers in 2030 is a great way to avoid admitting that the industry’s ‘senior-only’ hiring culture is the actual root of the rot. <a href="https://app.daily.dev/posts/jfqjRIR0E" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
