<h2>Must know</h2>
<ul>
<li><strong>Chrome zero-day under active exploitation</strong>: Google just patched a critical use-after-free vulnerability, CVE-2026-2441, in Chrome’s CSS font feature parsing that’s already being actively exploited. This is Chrome’s first zero-day of the year, and it allows for arbitrary code execution within the browser sandbox. The fix involves preventing modification of the underlying map during traversal, a class of bug that Rust’s ownership model would prevent, as noted by some technical analyses. <a href="https://app.daily.dev/posts/ISqaRw1RF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft Copilot chat leaked confidential emails</strong>: Microsoft confirmed a security bug in Copilot Chat (CW1226324) that allowed the AI to bypass confidentiality labels and Data Loss Prevention policies. It summarized emails from users’ Sent Items and Drafts folders without authorization, a flaw active since late January. This incident adds to growing enterprise anxiety about AI tools accessing sensitive corporate data, with Microsoft rolling out a fix in early February. <a href="https://app.daily.dev/posts/jSmIdhAQU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical vulnerabilities in popular VS Code extensions</strong>: OX Security disclosed critical vulnerabilities in four popular VS Code extensions—Live Server, Code Runner, Markdown Preview Enhanced, and Microsoft’s Live Preview—totaling 128 million installs. These flaws enable file theft, remote code execution, and local network reconnaissance. While Microsoft silently patched Live Preview, three other maintainers never responded, highlighting significant supply chain risks in developer tooling. <a href="https://app.daily.dev/posts/IuDICegHN" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI agent development and security accelerate</strong>: Enterprises are rapidly adopting and developing agentic AI, with new tools like StrongDM ID providing identity for AI agents and Cencurity offering security gateways to proxy LLM traffic, addressing governance and data leakage concerns. <a href="https://app.daily.dev/feed-by-ids?id=2UmMyMAEf&amp;id=32FCqrqN0&amp;id=4fmmBjp20&amp;id=Ck3BiV0To&amp;id=bbSNghGeS&amp;id=fvUXvxfty&amp;id=hYwz99Mp8&amp;id=i8naoDZlv&amp;id=nicRNl9Cx&amp;id=pJKpsRETp" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI model performance, cost, and sovereign AI initiatives</strong>: Anthropic’s new Claude Sonnet 4.6 is marketed as cheaper, but independent benchmarks suggest it’s a ‘token muncher’ due to adaptive thinking, potentially negating cost savings, while India and Japan push for sovereign AI with NVIDIA. <a href="https://app.daily.dev/feed-by-ids?id=i4O6sQBIw&amp;id=j12nF8mSg&amp;id=jj2Kauaop&amp;id=lUz1VvbAr&amp;id=y0N7mzotz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI’s impact on software development and engineering roles</strong>: Spotify’s CEO claims senior developers haven’t written code since December 2025, supervising AI instead, while Martin Fowler warns AI can accelerate technical debt if foundational practices are ignored. <a href="https://app.daily.dev/feed-by-ids?id=7hHra6Ai0&amp;id=WUJRo6v7h&amp;id=enHqyOYiB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-assisted coding tools and agentic development trends</strong>: New tools like DiffPrism are emerging for reviewing AI-generated code, while experts like José Valim advocate for integrated custom runtimes over fragmented MCP tooling for coding agents. <a href="https://app.daily.dev/feed-by-ids?id=XVyPsqBUP&amp;id=qulPHX7Sg&amp;id=zU6J5TUqH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 7.0 performance regressions on Intel panther lake</strong>: Early benchmarks show Linux 7.0 has performance regressions on Intel’s new Core Ultra Series 3 ‘Panther Lake’ hardware, affecting both CPU and iGPU performance compared to Linux 6.19. <a href="https://app.daily.dev/posts/5CIXyHDbR" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Tailscale peer relays general availability</strong>: Tailscale has made Peer Relays generally available, allowing users to deploy high-throughput relay nodes on any Tailscale device to improve network connectivity and performance, even in restrictive cloud environments. <a href="https://app.daily.dev/posts/Q0uzlfbeJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA run:ai GPU fractioning for LLM inference performance</strong>: NVIDIA and Nebius AI Cloud benchmarked GPU fractioning with NVIDIA Run:ai, showing that 0.5 GPU fractions deliver 77% of full-GPU throughput and 86% of concurrent user capacity for LLM inference, maximizing GPU ROI. <a href="https://app.daily.dev/posts/LsiRfCVWP" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI productivity impact: much ado about nothing?</strong>: An NBER survey of nearly 6,000 executives found over 80% reported no measurable impact from AI on productivity or employment over the past three years. Despite 69% of businesses using some form of AI, 89% saw no productivity gains. So, all that hype about AI revolutionizing the workplace? Turns out it’s mostly just… hype, for now. <a href="https://app.daily.dev/posts/rfmYnNc4e" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s C compiler: a marketing masterpiece of misleading claims</strong>: Anthropic released a marketing video claiming Claude built a C compiler ‘from scratch’ with ‘no human intervention’ that could compile Linux. Turns out, it used 37 years of GCC test suites, an online GCC oracle, required human restarts, cost $20,000 in API calls, and couldn’t even boot Linux. A genuine achievement of 16 AI agents running for two weeks was overshadowed by what can only be described as ‘dishonest marketing language.’ <a href="https://app.daily.dev/posts/0csgEkRMy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Mandatory AI tooling metrics: measuring activity, not outcomes</strong>: A developer is critiquing their company’s new internal metric mandating 80% of epics use AI tooling, arguing it measures tool usage rather than actual business outcomes. Because, clearly, the path to innovation is forcing everyone to use AI, regardless of whether it actually helps or just generates more ‘AI slop.’ What could go wrong when management prioritizes tool adoption over actual value? <a href="https://app.daily.dev/posts/8WhnSQAtX" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
