<h2>Must know</h2>
<ul>
<li><strong>AI coding agents are fast but flawed</strong>: AI coding agents like Cursor and Claude Code can accelerate development, but they often produce suboptimal, insecure, and hard-to-maintain code. Studies show AI-generated code has more bugs and security vulnerabilities, increasing review time and technical debt. While tools like Cursor’s Bugbot and Anthropic’s code-simplifier aim to improve quality, developers still need deep architectural understanding and rigorous human review to prevent issues, as highlighted by InfoWorld and The New Stack. The bottleneck is shifting from writing code to defining requirements and ensuring quality. <a href="https://app.daily.dev/feed-by-ids?id=jSgoQLSoF&amp;id=H430dwVHX&amp;id=J24eplQdr&amp;id=5khP9Wt37&amp;id=O3swuQWEt&amp;id=Oj1MubZPP&amp;id=QjH25jlip&amp;id=0Dxh8ahUn&amp;id=5NyDqJTPW&amp;id=6k9ppA468" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent security is a growing nightmare</strong>: AI agents and their underlying Model Context Protocol (MCP) servers are introducing critical security vulnerabilities. Google Gemini was exploited via prompt injection in calendar invites, and ServiceNow patched a critical flaw allowing unauthenticated user impersonation through AI agents. Anthropic’s Claude Code and Cowork also suffer from unpatched prompt injection vulnerabilities that can exfiltrate sensitive files. These incidents highlight that traditional security measures are failing against semantic attacks and the need for new approaches to govern LLMs and track data provenance, as reported by The Hacker News and InfoQ. <a href="https://app.daily.dev/feed-by-ids?id=dHjNBMkFO&amp;id=wzAqd15ET&amp;id=s11rN52dz&amp;id=jaKSkIh92&amp;id=rAHVvRPEj&amp;id=U6A8HrCeq&amp;id=hOkZZEdlQ&amp;id=j6bBmLvCW&amp;id=kZzcz2YmW&amp;id=mZZUloj80" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft open sources XAML Studio and patches critical vulnerabilities</strong>: Microsoft has open-sourced XAML Studio under the .NET Foundation, aiming to foster community collaboration for rapid UI prototyping. This move is intended to broaden support for XAML-based development beyond Microsoft platforms. Meanwhile, Microsoft addressed 112-114 vulnerabilities in its January Patch Tuesday, including a zero-day in Desktop Window Manager (CVE-2026-20805) actively exploited in the wild, and critical remote code execution bugs in Office. These updates underscore Microsoft’s dual focus on open-source engagement and urgent security remediation, as detailed by Krebs on Security and The Hacker News. <a href="https://app.daily.dev/feed-by-ids?id=hUJrXK08w&amp;id=Kwy6WAbQP" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI industry faces financial sustainability challenges</strong>: Despite massive investments, AI projects have a 90% failure rate, with 95% of enterprises reporting no ROI, leading to significant losses for companies like OpenAI and raising concerns about an AI industry bubble. <a href="https://app.daily.dev/feed-by-ids?id=DQh9IAXNl&amp;id=TNaBNlpBz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New open-source AI coding models emerge</strong>: GLM-4.7-Flash, a 30B-A3B Mixture of Experts model, is released for local AI coding and agentic tasks, outperforming peers in benchmarks, while open-weight models like GLM 4.7 and MiniMax M2.1 are nearing performance parity with frontier closed models at lower costs. <a href="https://app.daily.dev/feed-by-ids?id=eWdzokgG0&amp;id=p3khbP8jU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Small language models gain enterprise traction</strong>: Enterprises are increasingly favoring Small Language Models (SLMs) over LLMs due to lower costs, reduced latency, and better fit for specific tasks like invoice parsing and contract classification, often outperforming larger models in constrained business contexts. <a href="https://app.daily.dev/feed-by-ids?id=HX43N3MkE&amp;id=J1BKrq2u6" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA and Lilly launch $1B AI lab for drug discovery</strong>: NVIDIA and Eli Lilly are investing $1 billion in an AI-focused co-innovation lab to accelerate drug discovery by combining computational modeling with experimental wet labs, leveraging NVIDIA’s advanced AI infrastructure and BioNeMo platform. <a href="https://app.daily.dev/posts/TcHQkCPbq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Mandiant publishes NTLMv1 rainbow tables to force deprecation</strong>: Google-owned Mandiant released comprehensive rainbow tables for Net-NTLMv1, enabling credential recovery in under 12 hours with consumer hardware, aiming to force organizations to abandon the 20-year-old insecure protocol still prevalent in legacy systems. <a href="https://app.daily.dev/posts/2q2phjvWS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Node.js and Ruby release security and bugfix updates</strong>: Node.js released versions 20.20.0 LTS and 25.3.0, addressing multiple CVEs including a critical DoS vulnerability (CVE-2025-59466) affecting async_hooks, while Ruby 4.0.1 provides bugfixes, both focusing on stability and security. <a href="https://app.daily.dev/feed-by-ids?id=DwsoMgMKH&amp;id=WX4br4Zyw&amp;id=Yy5NFIIOM&amp;id=Zr0Dv9dwZ&amp;id=EyLXGRUe2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Crystal 1.19.0 advances multithreading and ARM64 support</strong>: Crystal 1.19.0 introduces execution contexts as a preview feature towards default multithreading in version 1.20, adds ARM64 Linux binaries, and enhances synchronization primitives, improving performance and cross-platform compatibility. <a href="https://app.daily.dev/posts/QEQMgHoGA" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI coding assistants: the emperor’s new code?</strong>: Apparently, AI coding assistants are making developers ‘slower’ and producing ‘low-quality’ code, according to some. Yet, Cursor is out there claiming its agents built a web browser from scratch in Rust, generating over a million lines of code in a week. The only catch? The code doesn’t compile, and there’s no working demo. So, are we just supposed to take their word for it, or is this another case of ‘meaningful progress’ that means absolutely nothing in the real world? <a href="https://app.daily.dev/feed-by-ids?id=V3KHuMgUU&amp;id=GPLZOgOgP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA’s CEO: ‘Doomer narrative’ or just inconvenient truths?</strong>: NVIDIA’s CEO, Jensen Huang, is out there calling AI skeptics ‘doomers’ for daring to suggest regulation or point out potential job displacement. He questions the motivations of ‘well-credentialed figures’ who warn about AI’s risks, implying they’re just trying to slow down innovation. Funny how the guy whose company is making billions from the AI boom thinks any criticism is just a ‘doomer narrative.’ Maybe it’s not a doomer narrative, Jensen, maybe it’s just people asking for a bit of accountability before we all get replaced by a chatbot. <a href="https://app.daily.dev/feed-by-ids?id=1vSR1HgdF&amp;id=JJAsCgIZe" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Enterprise AI: all hype, no ROI?</strong>: Gartner predicts global AI spending will hit $2.52 trillion by 2026, but MIT research says 95% of enterprises get zero ROI from their AI investments. So, we’re pouring trillions into something that mostly fails, and companies like OpenAI are losing billions. But don’t worry, vendors are just ‘absorbing short-term losses’ to ‘protect market share.’ Sounds like a solid business plan, right? Maybe we should just call it the ‘Trough of Disillusionment’ and move on. <a href="https://app.daily.dev/feed-by-ids?id=TNaBNlpBz&amp;id=DQh9IAXNl" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
