<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 active implementation of post-quantum security measures. They’ve released initial post-quantum cryptography standards and recommend hybrid cryptographic methods as interim solutions, emphasizing immediate action for comprehensive security. This transition is framed as urgent but long-term, addressing the risk of ‘harvest now, decrypt later’ attacks, as reported by multiple government sources. <a href="https://app.daily.dev/posts/lRZsyJgWP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical ServiceNow AI agent vulnerability disclosed</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 confirmed the vulnerability and released a patch, but the incident highlights the increased attack surface from rapid AI agent adoption, according to AppOmni. <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, aiming to correlate endpoint data with browser activity and address threats like session hijacking and data 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>Open-source PowerVR graphics drivers for BeaglePlay</strong>: The BeaglePlay board now supports fully open-source, upstream PowerVR graphics drivers, addressing long-standing demands from the Linux community for transparency and compatibility. <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>AI agents port Rust to Go with minimal human intervention</strong>: An experienced developer successfully ported a Rust-based Jinja2 template engine to native Go using AI agents, requiring only 45 minutes of active guidance and 10 hours of autonomous agent work, highlighting AI’s growing capability in cross-language porting. <a href="https://app.daily.dev/posts/VOMBPE1Xs" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Dragonfly graduates to full CNCF project maturity</strong>: Dragonfly, a cloud-native peer-to-peer image and file distribution platform, has graduated to full project maturity within the CNCF, demonstrating widespread adoption for container images, OCI artifacts, and AI models at scale. <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>Firefox 147 enhances privacy and performance</strong>: Firefox 147 introduces automatic Picture-in-Picture, 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>Intel Compute Runtime adds Crescent Island and Nova Lake S support</strong>: Intel’s Compute Runtime version 26.01.36711.4 introduces initial support for Crescent Island and Nova Lake S graphics hardware, maintaining Level Zero and OpenCL support for new and past Intel GPUs. <a href="https://app.daily.dev/posts/54Uqeadiv" 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 centralized Apache Spark-based system called FINUDP, aiming to eliminate metric inconsistencies and process millions of booking transactions daily with hourly updates. <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>Python’s cryptography library considers dropping OpenSSL support</strong>: Maintainers of Python’s cryptography library are expressing growing concerns over OpenSSL 3.0’s performance regressions and API complexity, considering a migration to alternative forks or dropping OpenSSL support entirely. <a href="https://app.daily.dev/posts/Wy77IQs4s" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI solves Erdős problems, but still needs human babysitting</strong>: OpenAI’s GPT 5.2 is apparently solving 11 of 15 newly resolved Erdős problems, and ChatGPT is now a ‘reasoning assistant’ for mathematicians. But don’t worry, Terence Tao says AI only excels at ‘obscure, straightforward’ problems, and humans still need to ‘verify proofs.’ So, AI is brilliant, but also too dumb to trust. Got it. <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>Anthropic’s Claude Cowork: great for enterprise, terrible for security</strong>: Anthropic released Cowork, a new feature for Claude Code, to automate enterprise tasks, but it’s already vulnerable to file exfiltration via indirect prompt injection. Anthropic’s response? ‘Users should monitor for suspicious activity.’ So, they’re selling advanced AI agents for enterprise, but the security model is ‘good luck, don’t get hacked.’ Classic. <a href="https://app.daily.dev/feed-by-ids?id=GMZHoSF60&amp;id=QvZHdydfi&amp;id=U2HzInRUP&amp;id=ejPwSyHLL&amp;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, apparently</strong>: Researchers supposedly replicated IBM quantum factorization results using a dog, suggesting IBM’s achievements are ‘overhyped.’ Meanwhile, Rice University is doing actual quantum simulator work on molecular energy transfer. It’s a bold strategy to critique quantum progress with canine-based experiments, but hey, anything for a headline, right? <a href="https://app.daily.dev/feed-by-ids?id=7Bd1YG4JM&amp;id=FBqZ5fVz5" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
