<h2>Must know</h2>
<ul>
<li><strong>PackageGate zero-days expose npm and yarn to worm attacks</strong>: Six zero-day vulnerabilities, dubbed PackageGate, are actively affecting npm and yarn, allowing attackers to bypass security controls designed to prevent worm attacks. While pnpm, vlt, and Bun have already patched these flaws, npm claims the behavior is ‘by design,’ leaving millions of JavaScript developers vulnerable. Oren Yomtov, the researcher who disclosed the vulnerabilities, strongly advises migrating to patched package managers immediately. <a href="https://app.daily.dev/posts/klUkuFZvK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft Office zero-day forces emergency patch</strong>: Microsoft issued an emergency out-of-band patch for a critical zero-day (CVE-2026-21509) in Microsoft Office, actively exploited in the wild. This flaw allows attackers to bypass security protections using malicious Office files. While newer versions are patched, Office 2016 and 2019 users are left with manual registry workarounds. CISA has mandated all federal agencies patch this vulnerability by February 16, 2026, highlighting the urgency of the threat. <a href="https://app.daily.dev/posts/qC937DXHe" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI use accelerates decline of public knowledge platforms</strong>: AI models like ChatGPT and Claude are causing a significant drop in traffic to public knowledge platforms such as Stack Overflow (down 78%) and Wikipedia. This shift risks a ‘model collapse’ where future AI models, trained on increasingly AI-generated content, degrade in quality due to detachment from human expertise. The trend fragments open collaboration and collective learning, favoring private, immediate problem-solving over communal knowledge accumulation, as reported by The Guardian. <a href="https://app.daily.dev/posts/3Qlqmk7FI" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI agents reshape developer roles and enterprise operations</strong>: AI agents are shifting developer roles from coding to product architecture and outcome engineering, automating much of the execution and creating new infrastructure needs for orchestration and fault tolerance, according to Temporal and ‘How Agents are quietly transforming the Enterprise’. <a href="https://app.daily.dev/feed-by-ids?id=b2fEANN64&amp;id=yBxtGCVqr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cloudflare launches serverless post-quantum Matrix homeserver</strong>: Cloudflare has ported a full Matrix homeserver, Tuwunel, to its Workers platform, leveraging serverless architecture and X25519MLKEM768 post-quantum cryptography to protect against ‘harvest-now-decrypt-later’ attacks, with idle servers costing less than $1 per month. <a href="https://app.daily.dev/posts/vURH9h7xw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Go language team proposes generic methods</strong>: The Go team has proposed adding generic methods to the language, allowing type parameters on concrete methods for improved code organization and readability, with full backward compatibility and static compile-time resolution. <a href="https://app.daily.dev/posts/W8OOKuC86" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Lennart Poettering founds Amutable for Linux integrity</strong>: Lennart Poettering, creator of systemd, has founded Amutable, a new company focused on bringing proactive, cryptographically verifiable integrity to Linux workloads, aiming to move infrastructure security beyond reactive, heuristic-based methods. <a href="https://app.daily.dev/posts/MvN4zVQYR" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>CISA publishes federal guidance on post-quantum cryptography adoption</strong>: CISA has released guidance mapping NIST-standardized post-quantum cryptography (PQC) algorithms to enterprise hardware and software, directing federal agencies to procure PQC-capable products where solutions are available, highlighting adoption gaps in digital signatures and authentication. <a href="https://app.daily.dev/posts/5Lj7cpvbX" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS Network Firewall adds GenAI traffic controls</strong>: AWS Network Firewall now supports web category-based filtering for generative AI application traffic, allowing security teams to block or restrict access to GenAI services at the network level and enforce compliance policies. <a href="https://app.daily.dev/posts/AdbvgaiZY" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ESLint v10.0.0-rc.2 released for testing</strong>: ESLint v10.0.0-rc.2, a release candidate, is now available, fixing a regression in the strict rule’s global mode report range and improving documentation, with a migration guide provided for upgrading. <a href="https://app.daily.dev/posts/EfrDMwvxT" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>D-Wave’s quantum computing roadmap: more hype, less substance</strong>: D-Wave is touting ‘314% year-over-year growth’ in Advantage2 usage and a $10 million contract, while simultaneously outlining an ‘accelerated gate-model quantum computing roadmap’ to launch its first system in 2026. Apparently, they’re the ‘only company developing all three core technologies’ for error-corrected systems, according to D-Wave. Funny how these self-reported milestones always seem to come directly from their own press releases, with zero independent verification. Sounds like a lot of future promises built on a foundation of ‘trust us, bro.’ <a href="https://app.daily.dev/posts/B9UXF3b8Z" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ricursive Intelligence: the $4 billion startup that materialized from thin air</strong>: Ricursive Intelligence, founded by ‘former Google researchers,’ just bagged $300 million at a $4 billion valuation, two months after launch. Their grand plan? An AI system to design and ‘automatically improve AI chips.’ The articles mention their founders ‘previously developed AlphaChip, reportedly used in Google’s TPU chips,’ which is a nice touch for a claim that’s completely unverified. No products, no public statements from founders, just a lot of money and a vague promise to make AI chips better with more AI. What could possibly go wrong? <a href="https://app.daily.dev/posts/pqGutcjN8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ChatGPT’s Grokipedia habit: when AI starts citing itself</strong>: ChatGPT’s GPT-5.2 model has been caught citing ‘Grokipedia,’ Elon Musk’s AI encyclopedia, in its responses. The Guardian found nine instances where ChatGPT referenced this AI-generated source, raising concerns about feedback loops amplifying biases and misinformation. OpenAI claims to have filters for low-credibility sources, but apparently, an AI citing another AI is perfectly fine. It’s like a snake eating its own tail, but with more confident, potentially inaccurate information. <a href="https://app.daily.dev/posts/fWRmbZwbT" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
