<h2>Must know</h2>
<ul>
<li><strong>AI coding agents are fast, but still need adult supervision</strong>: AI coding agents like Claude Code are showing impressive speed, generating millions of lines of code in days for projects like the FastRender web browser. However, they still struggle with ambiguous requirements, often resort to code repetition, and introduce vulnerabilities, according to developer accounts and research cited in TechCrunch. GitHub Copilot users, for instance, see a 322% increase in code vulnerabilities. The consensus is that while AI can accelerate development, human oversight, strong testing, and clear architectural guidance remain critical to prevent technical debt and quality regressions. <a href="https://app.daily.dev/feed-by-ids?id=EqMZfdLoq&amp;id=GPLZOgOgP&amp;id=J24eplQdr&amp;id=cSbwL1BS5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes is now the enterprise AI standard, culture is the new bottleneck</strong>: Kubernetes is now used in production by 82% of organizations, up from 66% in 2023, solidifying its role as the standard for enterprise workloads, especially AI inference (66% of AI adopters use it). The 2025 CNCF Annual Cloud Native Survey, widely cited, indicates that cultural resistance, not technical complexity, is now the primary barrier to adoption. Mature organizations are leveraging GitOps and platform engineering, but the shift highlights that people, not just tech, need to evolve. <a href="https://app.daily.dev/posts/Lwsgn3lZ5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft mandates Intune security updates or your apps get blocked</strong>: Starting January 19, 2026, Microsoft is enforcing mandatory Intune Mobile Application Management security updates. Enterprises must update iOS and Android apps to the latest Intune SDK versions, or risk having them blocked from accessing corporate resources like Outlook and Teams. This move introduces new security controls like PIN requirements and biometric authentication, but offers no grace period, meaning non-compliant apps will immediately cease to function. <a href="https://app.daily.dev/posts/PNEoUht10" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>North Korean hackers target fintech developers via VS Code</strong>: North Korean threat actors are exploiting VS Code project files to deploy malware and backdoors on developer machines, specifically targeting those in cryptocurrency, blockchain, and fintech sectors. <a href="https://app.daily.dev/feed-by-ids?id=1wKLyk8gG&amp;id=Q9M3WuQfH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>EU launches decentralized vulnerability database</strong>: The EU has launched <a href="http://db.gcve.eu" target="_blank" rel="noopener nofollow">db.gcve.eu</a>, a public cybersecurity vulnerability database, aiming to reduce reliance on US systems by integrating data from over 25 sources and enabling autonomous vulnerability ID assignment. <a href="https://app.daily.dev/posts/i2uCJ3RGE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s Git MCP server had critical vulnerabilities</strong>: Anthropic’s Git MCP server was found to have three critical vulnerabilities, including path validation bypass and argument injection, which could have allowed remote code execution via prompt injection. <a href="https://app.daily.dev/posts/U6A8HrCeq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Taiwan and US ink AI and semiconductor supply chain pact</strong>: Taiwan and the US have finalized a strategic trade agreement focused on semiconductors and AI supply chains, with Taiwan committing $250 billion in investments and receiving lowered tariffs. <a href="https://app.daily.dev/posts/suaVSewS2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New GLM-4.7 Flash model excels in local AI coding</strong>: GLM-4.7 Flash, a 30B-A3B mixture-of-experts model, reportedly outperforms competitors like Qwen 3-30B-A3B in coding and agentic AI benchmarks, offering superior tool calling and fast inference for local AI tasks. <a href="https://app.daily.dev/posts/eWdzokgG0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>One Identity Manager 10.0 enhances security with AI-assisted reporting</strong>: One Identity has released Identity Manager 10.0, featuring risk-based governance, identity threat detection and response (ITDR) capabilities, and AI-assisted reporting via natural language querying. <a href="https://app.daily.dev/posts/xsUICDLZg" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft updates React Native for Windows, still lags mobile</strong>: Microsoft released React Native Windows 0.81 with Hermes debugger support and Fabric architecture, but it still lags behind mobile versions and has limited third-party package support. <a href="https://app.daily.dev/feed-by-ids?id=7jYV4dEG1&amp;id=qcm6RjfK7" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI investment payoff still a mystery for most CEOs</strong>: PwC’s latest survey of 4,454 CEOs reveals a shocking truth: 56% haven’t seen any revenue increase or cost decrease from their AI spending. Only 12% reported both. So much for the AI revolution, looks like most are still waiting for the magic to happen, or perhaps, for the consultants to explain why it hasn’t. <a href="https://app.daily.dev/feed-by-ids?id=Kov24DV15&amp;id=ogb2Wd6xs&amp;id=rEdiBiFbV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vibecoding: the new executive superpower (or just more AI hype)?</strong>: Emergent, an Indian AI startup, just bagged $70 million for ‘vibecoding,’ a term that apparently means building apps with AI agents without needing ‘advanced coding expertise.’ An opinion piece suggests executives should embrace this to become ‘individual contributors.’ Because nothing says ‘cutting-edge tech’ like a buzzword that sounds like a bad TikTok trend, promising to turn every CEO into a developer overnight. <a href="https://app.daily.dev/feed-by-ids?id=AllrxKqHW&amp;id=McILqauKu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code: the gift that keeps on leaking</strong>: Turns out, AI-generated SwiftUI code is a master of creating persistent memory leaks by strongly capturing <code>self</code> in nested closures. Engineers had to manually refactor the AI’s ‘brilliant’ work with weak references and explicit capture lists. So, while AI can churn out code, it seems it still needs humans to clean up its messes and teach it basic memory management, proving that some ‘best practices’ are still best left to actual practitioners. <a href="https://app.daily.dev/feed-by-ids?id=0Dxh8ahUn&amp;id=NOlmy6yd2&amp;id=yTmR3XbQ9" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
