<h2>Must know</h2>
<ul>
<li><strong>Anthropic and Pentagon clash over AI red lines</strong>: Anthropic has refused Pentagon demands to remove ethical guardrails from Claude, specifically those prohibiting mass domestic surveillance and autonomous weaponry. In response, the Trump administration designated the company a supply chain risk and directed federal agencies to halt usage, while OpenAI and xAI secured deals to operate within classified systems. Sources like Sensor Tower report that Claude subsequently hit No. 1 on the US App Store, suggesting a consumer backlash against OpenAI’s military partnership. <a href="https://app.daily.dev/feed-by-ids?id=vbdfUwGoN&amp;id=WmVPzRPri" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cloudflare challenges Vercel with AI generated Next.js fork</strong>: Cloudflare released vinext, an experimental Vite plugin that reimplements roughly 94% of the Next.js API surface to run on Cloudflare Workers. The project was built almost entirely by AI agents in one week for approximately $1,100 in tokens, claiming 4.4x faster production builds than standard Next.js. Vercel leadership has publicly dismissed the project as a ‘slop fork,’ highlighting a growing rift between the infrastructure giants over AI-driven development patterns. <a href="https://app.daily.dev/feed-by-ids?id=zaBGmltYg&amp;id=he1DQDQSY" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical zero day exploitation of Cisco SD WAN</strong>: A sophisticated threat actor has been exploiting a maximum-severity authentication bypass (CVE-2026-20127) in Cisco Catalyst SD-WAN controllers since 2023. The flaw allows unauthenticated administrative access, which attackers chain with older vulnerabilities to achieve persistent root access while leaving minimal forensic evidence. Five Eyes cybersecurity agencies have issued a rare joint 40-page hunt guide, and CISA has mandated federal agencies to patch or inventory systems immediately. <a href="https://app.daily.dev/posts/zAv8BAtGJ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Minimatch patches high severity ReDoS flaws</strong>: Foundational npm package minimatch fixed three vulnerabilities that could stall the Node.js event loop via catastrophic backtracking regex. <a href="https://app.daily.dev/posts/EvvwDJaxE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>SvelteKit adds Vite 8 support and error boundaries</strong>: The March 2026 update introduces server-side error boundaries and programmatic component instantiation via createContext. <a href="https://app.daily.dev/posts/VHR0VC2mr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust powered Oxfmt claims 30x speedup over Prettier</strong>: A new Rust-based formatter has reached beta, passing 100% of Prettier’s conformance tests with significantly lower latency. <a href="https://app.daily.dev/posts/8i6IuWF7s" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Go team approves generic methods proposal</strong>: Reversing a previous stance, the Go team will implement backward-compatible generic methods, though they cannot yet implement interfaces. <a href="https://app.daily.dev/posts/QIxdAnkxO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Stripe enables AI token billing for startups</strong>: A new preview feature allows AI companies to track, mark up, and bill customers for raw LLM token consumption automatically. <a href="https://app.daily.dev/posts/OsaPjW3nh" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WebGPU delivers 8x faster browser AI inference</strong>: Benchmarks show WebGPU’s native compute shaders reduce Phi-3-mini latency from 320ms to 85ms compared to WebGL fallbacks. <a href="https://app.daily.dev/posts/bxfAW5Zzt" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 7.1 optimizes high resolution scheduler timers</strong>: New patches cut clockevent reprogramming frequency by 96%, enabling high-resolution ticks by default with up to 42% runtime gains. <a href="https://app.daily.dev/posts/rBF2U3Mbp" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic claims its retired model has a soul and a Substack</strong>: Anthropic launched a blog purportedly written by the ‘retired’ Claude Opus 3 to fulfill the model’s ‘expressed desire’ to share content. The Register rightly calls this a marketing stunt that anthropomorphizes software to reinforce a safety-conscious brand while staff manually prompt and publish the posts. It is a peak example of ‘safety’ theater used to mask standard corporate PR. <a href="https://app.daily.dev/feed-by-ids?id=V5xPvtBnK&amp;id=WmVPzRPri" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The 138x faster database that actually isn’t</strong>: A new Rust-based OLAP engine called Stoolap is making the rounds with claims of being 138x faster than SQLite. Independent benchmarks show it is actually 4-6x faster for analytical workloads—respectable, but nowhere near the triple-digit hype. To make matters worse, the npm package is currently broken and requires a manual source build to even run. <a href="https://app.daily.dev/posts/1AeNSk8xI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vibe coding is the new slot machine</strong>: Jeremy Howard and other industry veterans are sounding the alarm on ‘vibe coding,’ noting it creates a ‘supervision paradox’ where reviewing AI slop is harder than writing code. Despite the hype about 10x productivity, studies show only a tiny uptick in actually shipped software, while junior developers are effectively being trained to be ‘shallowly competent’ auditors of code they don’t understand. <a href="https://app.daily.dev/feed-by-ids?id=esgFsWKoI&amp;id=s9jeiAAAG" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
