<h2>Must know</h2>
<ul>
<li><strong>AI is getting scary good at making malware and exploits</strong>: New research shows AI models like GPT-5.2 and Opus 4.5 can generate sophisticated zero-day exploits, bypassing modern security measures, with one exploit costing only $50 in compute. Check Point Research also identified VoidLink, the first AI-generated malware suite, developed in a week by a single person using AI agents—a task that would take human teams 30 weeks. This means the limiting factor for exploit generation is now AI token throughput, not human expertise, according to WkT9zuzY1 and nLTC7GTyu. <a href="https://app.daily.dev/feed-by-ids?id=WkT9zuzY1&amp;id=nLTC7GTyu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes is now the standard for enterprise AI, but culture is the new bottleneck</strong>: The 2025 CNCF survey reveals Kubernetes is used by 82% of organizations in production, with 66% using it for AI inference. While cloud-native adoption is nearly universal, cultural resistance, not technical hurdles, is now the primary barrier to further adoption. Mature organizations are leveraging GitOps and platform engineering, but the survey also flags sustainability concerns for open-source infrastructure due to increasing AI workloads, as reported by Lwsgn3lZ5. <a href="https://app.daily.dev/posts/Lwsgn3lZ5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Taiwan and US ink major AI and semiconductor deal</strong>: Taiwan and the United States have finalized a strategic trade agreement focused on securing AI and semiconductor supply chains. Taiwanese chipmakers, including TSMC, will expand US manufacturing, benefiting from reduced tariffs and duty-free export status for some items. Taiwan is committing $250 billion in investments and an additional $250 billion in credit guarantees for US-based tech projects, as detailed in suaVSewS2. <a href="https://app.daily.dev/posts/suaVSewS2" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI coding agents are fast, but still need adult supervision</strong>: AI coding agents like Claude Code can deliver significant velocity gains for well-defined tasks, but struggle with ambiguous requirements, often producing repetitive code and omitting tests, requiring substantial human oversight and detailed guidance. <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>Google Gemini prompt injection raises enterprise security alarms</strong>: Security researchers found a vulnerability in Google Gemini where malicious calendar invitations could exploit the model’s automatic data ingestion to manipulate its behavior and exfiltrate private meeting data, highlighting new prompt injection risks in LLM-based systems. <a href="https://app.daily.dev/feed-by-ids?id=dHjNBMkFO&amp;id=5ZpMYMEZm&amp;id=dRAo2Mdf1&amp;id=j6bBmLvCW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft mandates Intune security updates for mobile apps</strong>: From January 19, 2026, Microsoft will enforce mandatory Intune Mobile Application Management security updates, requiring enterprises to update iOS and Android apps to the latest SDK versions or face immediate blocking from corporate resources. <a href="https://app.daily.dev/posts/PNEoUht10" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New EU vulnerability database aims for decentralization</strong>: The EU has launched <a href="http://db.gcve.eu" target="_blank" rel="noopener nofollow">db.gcve.eu</a>, a public, decentralized cybersecurity vulnerability database integrating data from over 25 sources, aiming to reduce reliance on US-based systems and provide autonomous vulnerability identifier assignment. <a href="https://app.daily.dev/posts/i2uCJ3RGE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>North Korean hackers target fintech developers via VS Code</strong>: North Korean threat actors are exploiting VS Code project files to deploy backdoors and malware on developer machines, targeting those in cryptocurrency, blockchain, and fintech sectors through fake job assessment projects. <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>Anthropic fixes critical Git server vulnerabilities</strong>: Anthropic patched three critical vulnerabilities in its Git MCP server that could have allowed remote code execution via prompt injection, highlighting security risks in AI-integrated systems that chain tools or components. <a href="https://app.daily.dev/posts/U6A8HrCeq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New tool streamlines cloud-native testing on Kubernetes</strong>: MetalBear released mirrord for CI, a new tool that allows CI tests to run directly on shared Kubernetes clusters, proxying traffic and environment variables to simulate production conditions without deploying code or setting up ephemeral environments. <a href="https://app.daily.dev/posts/s2plqXXTq" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI ‘vibecoding’ startup raises $70M, because who needs actual coding skills?</strong>: Emergent, an Indian AI-powered ‘vibecoding’ startup, just bagged $70 million in Series B funding, claiming $50 million ARR and 5 million users. Apparently, their platform lets you build full-stack apps with AI agents, completely sidestepping the need for ‘advanced coding expertise.’ An opinion piece even suggests executives should be ‘vibecoding’ to become individual contributors, because why bother with engineering staff when you have AI to do all the ‘vibes’ for you? Sounds like a recipe for some truly ‘vibey’ technical debt. <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>CEOs still waiting for AI to magically make money</strong>: A PwC survey of over 4,000 CEOs found that 56% haven’t seen any revenue increase or cost decrease from their AI investments. Only a measly 12% reported both. Meanwhile, only 14% of workers use generative AI daily. So, while the industry keeps pushing ‘agentic AI’ as the next big thing, most companies are still trying to figure out how to get a return on their initial AI hype-driven spending. Maybe they should try ‘vibecoding’ their balance sheets? <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>GenAI hype cycle: the gift that keeps on giving (disappointment)</strong>: It seems every major AI model release follows the same predictable pattern: exaggerated claims of revolution, followed by a quick dose of reality as limitations emerge, and finally, widespread disappointment. Social media fuels this cycle, pushing unrealistic predictions about job displacement, especially for software engineers. Analysts are now warning that anyone who only talks about the ‘benefits’ of GenAI without acknowledging the drawbacks is doing everyone a disservice. Who knew that ignoring reality leads to bad decisions? Shocking. <a href="https://app.daily.dev/feed-by-ids?id=j0xF0WiVr&amp;id=yCkgCoa6x" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
