<h2>Must know</h2>
<ul>
<li><strong>US agencies push post-quantum security transition</strong>: U.S. federal agencies, including the FBI, NIST, and CISA, are moving beyond awareness to actively implement post-quantum security measures. They’ve released initial cryptographic standards and recommend hybrid methods as interim solutions, emphasizing immediate action for comprehensive security against ‘harvest now, decrypt later’ threats. This transition is framed as urgent but long-term, requiring organizations to start cryptographic inventory and integration efforts now, as reported by various news outlets. <a href="https://app.daily.dev/posts/lRZsyJgWP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical ServiceNow AI agent vulnerability found</strong>: AppOmni discovered ‘BodySnatcher’ (CVE-2025-12420), a critical ServiceNow flaw allowing unauthenticated attackers to impersonate users with just an email, bypassing MFA and SSO. This could lead to rogue AI agents with escalated privileges accessing external environments. ServiceNow has patched the vulnerability, but the incident highlights the increased attack surface from rapid AI agent adoption, according to AppOmni’s disclosure. <a href="https://app.daily.dev/posts/wzAqd15ET" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>CrowdStrike acquires Seraphic Security for browser protection</strong>: CrowdStrike is acquiring Seraphic Security for approximately $400 million, marking its fourth acquisition since August. Seraphic specializes in enterprise browser security, using an abstraction layer to isolate JavaScript engines and randomize memory addresses against exploits. CrowdStrike plans to integrate Seraphic into its Falcon platform, combining endpoint data with browser activity for real-time threat correlation, addressing risks from session hijacking and exfiltration via AI prompts, as detailed in official announcements. <a href="https://app.daily.dev/posts/AKJAUyPZD" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Firefox 147 enhances privacy and performance</strong>: Firefox 147 introduces automatic Picture-in-Picture, reorganized tab settings, enhanced WebGPU support for Apple Silicon Macs, zero-copy hardware video playback for Windows AMD users, and XDG Base Directory support for Linux, alongside several security fixes and an upgrade to Google Safe Browsing v5. <a href="https://app.daily.dev/posts/f7pzHnDhN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>CNCF graduates Dragonfly for cloud-native distribution</strong>: Dragonfly, a peer-to-peer image and file distribution platform for container images and AI models, has graduated from CNCF incubation, demonstrating production readiness and widespread adoption by major organizations like Alibaba and Datadog. <a href="https://app.daily.dev/feed-by-ids?id=0rVfXzPjQ&amp;id=bjqQdB8fJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Livewire 4 simplifies component development</strong>: Livewire 4 introduces single-file components, an ‘islands’ feature for isolated rendering and performance improvements, built-in drag-and-drop, smooth transitions via View Transitions API, and expanded JavaScript interoperability, all while maintaining backward compatibility. <a href="https://app.daily.dev/posts/OxhgZzaoQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Agoda centralizes financial data with Apache Spark</strong>: Agoda consolidated its financial data pipelines into a unified Apache Spark-based platform, FINUDP, to eliminate metric inconsistencies, processing millions of transactions daily with multi-layered quality control and machine learning anomaly detection. <a href="https://app.daily.dev/feed-by-ids?id=SIYRILab7&amp;id=WuRILab7" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>BeaglePlay gets fully open-source PowerVR graphics</strong>: The BeaglePlay board now supports fully open-source, upstream PowerVR graphics drivers, addressing long-standing demands from the Linux community for transparency and improved compatibility with Imagination Technologies’ GPUs. <a href="https://app.daily.dev/feed-by-ids?id=HwIiwOf9Q&amp;id=NjVG9GX8g" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Python’s cryptography library considers dropping OpenSSL</strong>: Maintainers of Python’s cryptography library are concerned about OpenSSL’s performance regressions and API complexity, exploring migration to forks like BoringSSL or LibreSSL, and potentially dropping OpenSSL support entirely, having achieved 10x performance with their own Rust-based parsing. <a href="https://app.daily.dev/posts/Wy77IQs4s" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New Linux malware targets cloud environments</strong>: Security researchers discovered VoidLink, a sophisticated Linux malware framework with over 30 modules for reconnaissance and lateral movement, specifically targeting cloud platforms like AWS, GCP, and Azure, indicating a growing focus by professional threat actors on cloud infrastructure. <a href="https://app.daily.dev/posts/x9wl564Rc" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI solves math problems, but still needs human babysitters</strong>: OpenAI’s GPT 5.2 is apparently solving a bunch of obscure math problems, and tools like Harmonic’s Aristotle are helping with formalization. Mathematician Terence Tao says AI is great for the ‘long tail’ of straightforward problems. But don’t worry, humans still need to ‘verify proofs’ and handle the ‘complex/non-straightforward cases.’ So, AI is just doing the grunt work, leaving the real thinking to us. How revolutionary. <a href="https://app.daily.dev/feed-by-ids?id=8o6H9wDXV&amp;id=mhyD4ij4A" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI assistant vulnerabilities: because who needs security when you have ‘innovation’?</strong>: Microsoft Copilot apparently had a vulnerability that let attackers exfiltrate data with a single malicious URL click, and Claude Cowork is susceptible to indirect prompt injection for file exfiltration. Microsoft patched theirs, but Anthropic’s response to Cowork’s flaw is basically ‘users beware.’ It’s almost like rushing AI tools to market without robust security is a bad idea. Who could have predicted that? <a href="https://app.daily.dev/feed-by-ids?id=G7pZqTe0S&amp;id=jaKSkIh92&amp;id=rAHVvRPEj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Production apps need more than just code. Shocking!</strong>: Apparently, production applications require things like analytics, error tracking, uptime monitoring, rate limiting, and automated testing—features ‘tutorials never teach you.’ Who knew that building robust, reliable software involved more than just writing functional code? It’s almost as if there’s a difference between a hobby project and something that actually needs to work in the real world. Mind. Blown. <a href="https://app.daily.dev/feed-by-ids?id=hizCv12EP&amp;id=q1wpmdkFL" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
