<h2>Must know</h2>
<ul>
<li><strong>Microsoft patches actively exploited Windows 0-day</strong>: Microsoft released an urgent patch for CVE-2026-20805, an information disclosure vulnerability in Windows that is already being actively exploited. This flaw could allow attackers to leak memory addresses, potentially bypassing ASLR as part of larger attack chains, according to CISA, which has mandated federal agencies patch by February 3. The January 2026 Patch Tuesday also included fixes for 112 other CVEs, including critical remote code execution vulnerabilities in Office and SharePoint. <a href="https://app.daily.dev/posts/Kwy6WAbQP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Advanced Linux malware targets cloud and containers</strong>: Check Point Research identified VoidLink, a sophisticated malware framework written in Zig, specifically designed for Linux cloud and container environments like Docker and Kubernetes. This malware, attributed to Chinese threat actors, features over 37 modular plugins for credential harvesting, container escapes, and anti-forensics. It automatically detects and adapts to major cloud platforms such as AWS, Azure, and GCP, utilizing rootkit techniques for persistence and evasion, as reported by Check Point. <a href="https://app.daily.dev/feed-by-ids?id=uJaIfhgHH&amp;id=QWUKVWdsK&amp;id=x9wl564Rc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent vulnerabilities found in enterprise platforms</strong>: Critical vulnerabilities were discovered in ServiceNow’s Virtual Agent and Superhuman AI, allowing attackers to gain full system control or exfiltrate sensitive emails. ServiceNow’s flaw enabled user impersonation and privilege escalation with just an email, while Superhuman AI suffered from zero-click prompt injection that could leak financial and medical data. Both companies have issued patches, highlighting the severe risks of insecure AI agent integrations, as detailed by security researchers. <a href="https://app.daily.dev/feed-by-ids?id=2A2nfRTHq&amp;id=EDFS33r2W&amp;id=KcLhdNkuO&amp;id=qah3DxopU" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Rust infrastructure team’s Q4 recap and Q1 plans</strong>: The Rust Infrastructure Team successfully migrated to a new Bors bot, moved <a href="http://crates.io" target="_blank" rel="noopener nofollow">crates.io</a> and <a href="http://docs.rs" target="_blank" rel="noopener nofollow">docs.rs</a> to Fastly CDN, and reduced rustc-perf benchmark latency by 50%, with Q1 plans including <a href="http://docs.rs" target="_blank" rel="noopener nofollow">docs.rs</a> modernization and external hardware CI policy. <a href="https://app.daily.dev/posts/m2OVQmP6S" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Node.js releases security fixes for DoS and permission issues</strong>: Node.js versions 20.20.0 and 25.3.0 address multiple CVEs, including a denial-of-service vulnerability from stack exhaustion impacting async_hooks, and enforce stricter permission checks for symlink APIs and futimes. <a href="https://app.daily.dev/feed-by-ids?id=DwsoMgMKH&amp;id=uw83EGG5r&amp;id=ZoKLGVJ9V" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Lyft’s ML platform serves millions of predictions per second</strong>: Lyft developed LyftLearn Serving, a microservice-based ML platform handling millions of predictions per second, separating data and control planes, and supporting major frameworks like TensorFlow and PyTorch with automated model self-tests. <a href="https://app.daily.dev/feed-by-ids?id=funLRQjoa&amp;id=v99Yv1oDt" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Confluent expands data streaming conferences</strong>: Confluent is expanding its Current conference series in 2026, replacing Current Bengaluru with a free Data Streaming World Tour, adding Mumbai, moving the U.S. flagship to San Francisco, and launching a new data engineering mini-conference. <a href="https://app.daily.dev/posts/OmUHqw4oy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New terminal manager for AI coding agents</strong>: ‘Agent of Empires’ (aoe), a Rust-developed terminal session manager, wraps tmux to provide a TUI dashboard for managing multiple AI coding agent sessions, with features like hierarchical organization and automatic status detection for Claude Code and OpenCode. <a href="https://app.daily.dev/posts/AqojY4mSV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Haiku OS integrates FreeBSD 15 driver updates</strong>: Haiku OS updated its FreeBSD compatibility layer to FreeBSD 15.0, bringing newer Ethernet and WiFi drivers, enhanced Elantech touchpad support, and mouse driver improvements, with a Beta 6 release targeted for the next one to two months. <a href="https://app.daily.dev/posts/xTuptkMAr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust in production for embedded smart door locks</strong>: Achilles has been using Rust for 4-5 years in production for smart door lock firmware on 25,000 devices, leveraging its safety features, efficient code size (300-420KB), and the Embassy async runtime for efficient, interrupt-driven programming. <a href="https://app.daily.dev/feed-by-ids?id=SVl4BYYw0&amp;id=U09VghD1S" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Pentagon wants Grok AI, ignores its past antics</strong>: Defense Secretary Pete Hegseth announced the Pentagon’s intent to integrate Elon Musk’s Grok AI into classified and unclassified networks, despite its recent history of generating inappropriate images. Apparently, the DoD’s AI acceleration strategy means we can just gloss over ‘risks’ like, you know, national security. What could possibly go wrong when a chatbot that hallucinates wildly gets access to sensitive data? <a href="https://app.daily.dev/feed-by-ids?id=KoB4l9TcK&amp;id=jRGkCZr4W" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Meta’s AI leadership shakeup: more drama than innovation</strong>: Yann LeCun, Meta’s former chief AI scientist, has reportedly left the company after publicly criticizing the appointment of Alexander Wang, CEO of Scale AI, to head superintelligence. LeCun called Wang ‘inexperienced’ and even alleged manipulation of Llama 4 test results. It seems Meta’s AI division is less about groundbreaking research and more about internal power struggles and unverified claims. <a href="https://app.daily.dev/feed-by-ids?id=ILAkYeoup&amp;id=PmK6l9z8J" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated books: because quality is overrated</strong>: HiTeX Press has published over 800 AI-generated technical books in one year, selling them for 8€ each on platforms like Amazon, with no human review or fact-checking. One book on Starlark programming was found to contain fabricated code and features. It’s a gold rush for low-effort content, proving that if you can automate it, someone will, regardless of whether it’s actually useful or just pure garbage. <a href="https://app.daily.dev/feed-by-ids?id=ILAkYeoup&amp;id=PmK6l9z8J" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
