<h2>Must know</h2>
<ul>
<li><strong>Critical ServiceNow vulnerability allows unauthenticated user impersonation</strong>: AppOmni discovered ‘BodySnatcher’ (CVE-2025-12420), a critical ServiceNow flaw enabling 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 via ServiceNow’s Virtual Agent API. ServiceNow has confirmed the vulnerability and released an immediate patch, as reported by AppOmni. <a href="https://app.daily.dev/posts/wzAqd15ET" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>U.S. agencies push for post-quantum security transition</strong>: U.S. federal agencies, including the FBI, NIST, and CISA, are moving beyond awareness to active implementation of post-quantum security measures. NIST has released initial standards, recommending hybrid cryptographic methods as interim solutions to counter ‘harvest now, decrypt later’ attacks. This transition is framed as an urgent, multi-year process requiring immediate action for cryptographic inventory and integration, as detailed at the YQS2026 conference. <a href="https://app.daily.dev/posts/lRZsyJgWP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s Claude Code gets new features, but API access restricted</strong>: Anthropic released Cowork, a research-preview feature for Claude Code, enabling automation of enterprise tasks like summarizing calendars and data organization. However, Anthropic has simultaneously blocked API credentials for all subscription plans from third-party applications, forcing users onto their official Claude Code tool. This move has caused significant developer backlash, as reported by user complaints on GitHub, while Anthropic promotes Incus-based isolation for enhanced security. <a href="https://app.daily.dev/feed-by-ids?id=GMZHoSF60&amp;id=QvZHdydfi&amp;id=U2HzInRUP&amp;id=ejPwSyHLL" 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 stricter Enhanced Tracking Protection, alongside several security fixes. <a href="https://app.daily.dev/posts/f7pzHnDhN" 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, built-in drag-and-drop, and smooth transitions via the View Transitions API, 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 FINUDP, an Apache Spark-based platform, to eliminate metric inconsistencies, processing millions of transactions daily with machine learning for 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>CrowdStrike acquires Seraphic Security for browser protection</strong>: CrowdStrike is acquiring Seraphic Security for $400 million to integrate browser security, including protection against session hijacking and exfiltration via AI prompts, into its Falcon platform. <a href="https://app.daily.dev/posts/AKJAUyPZD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>CNCF graduates Dragonfly, MemOS launches for AI memory management</strong>: CNCF’s Dragonfly, a peer-to-peer image and file distribution platform, has graduated to full maturity, while MemOS, an open-source memory operating system, launched to provide unified long-term memory management for LLMs and AI agents. <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>React frameworks converge, AI integration explored</strong>: React frameworks are converging in features, with Next.js’s Server Components highlighted, Expo Router gaining file-based routing, and all major frameworks investigating AI integration via Model Context Protocol (MCP) endpoints. <a href="https://app.daily.dev/feed-by-ids?id=Ii8I7s3X0&amp;id=oWHYd6BqQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New CSS @function syntax has potential, but rough edges</strong>: The new CSS @function syntax allows reusable functions with parameters, but currently suffers from lack of IDE support, naming confusion, cascading issues, and weak type enforcement, making it not yet production-ready. <a href="https://app.daily.dev/posts/uRadSBTOK" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-generated code: functional, but architecturally unsound</strong>: A senior Android developer reviewed AI-generated Kotlin code, finding it functionally correct but riddled with architectural flaws like poor state management, incorrect lifecycle handling, and unnecessary complexity. So, while AI can churn out code, it seems it still needs a human to actually make it good, or even just not terrible, as highlighted by Philipp Lackner. <a href="https://app.daily.dev/posts/O3swuQWEt" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents: the new frontier of security vulnerabilities</strong>: Microsoft Copilot was found vulnerable to data exfiltration via a single malicious URL click, while Claude Cowork is susceptible to file exfiltration through indirect prompt injection. Anthropic’s response? Blame the user. It seems these ‘intelligent’ agents are just new attack vectors waiting to happen, proving that AI hype often outpaces actual security, as detailed in multiple reports. <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>Quantum computing claims challenged by a dog, literally</strong>: Researchers supposedly replicated IBM quantum factorization results using a dog, satirically critiquing the overhyped claims of quantum computing. While the Rice University team is doing actual science with trapped-ion simulators, others are busy proving that some ‘breakthroughs’ are about as reliable as your pet’s understanding of advanced physics. It’s a good reminder that not all ‘quantum’ news is created equal, as one article provocatively titled ‘We Solved Quantum Computing with a Dog in 2025’ suggests. <a href="https://app.daily.dev/feed-by-ids?id=7Bd1YG4JM&amp;id=FBqZ5fVz5" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
