<h2>Must know</h2>
<ul>
<li><strong>AMD CPUs vulnerable to StackWarp attack</strong>: A new hardware vulnerability, StackWarp (CVE-2025-29943), affects AMD Zen 1-5 CPUs, allowing privileged attackers to bypass SEV-SNP confidential VM protections. This exploit manipulates the CPU stack engine to hijack control flow without reading encrypted memory. AMD has released microcode updates, with more firmware patches expected, and advises disabling hyperthreading on critical systems as a mitigation. <a href="https://app.daily.dev/posts/9DURDnZX3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Malicious Chrome extensions steal enterprise session tokens</strong>: Five Chrome extensions, disguised as productivity tools, were found stealing authentication session tokens from enterprise HR and ERP platforms like Workday and NetSuite. These sophisticated attacks bypassed multi-factor authentication, allowing full account takeover. Socket.dev researchers reported the issue to Google, but thousands of users were already affected, highlighting a significant supply chain security risk. <a href="https://app.daily.dev/posts/3ImoRkfEG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI security and governance investments surge</strong>: CISOs are prioritizing AI-driven cybersecurity, focusing on identity access management for AI agents, enhanced email security against AI-generated phishing, and automated vulnerability discovery. Witness AI, a startup, secured $58 million to combat ‘shadow AI’ and compliance issues, reflecting venture capitalists’ projection of an $800 billion to $1.2 trillion AI security market by 2031. This indicates a critical shift in security priorities towards AI-specific threats. <a href="https://app.daily.dev/feed-by-ids?id=2XMeIQvIr&amp;id=6gbkOathj" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>GitLab 18.8 launches AI-orchestrated DevOps platform</strong>: GitLab 18.8 introduces the Duo Agent Platform, now generally available, enabling AI agents to orchestrate software development lifecycle tasks, including a Planner Agent for work items and a Security Analyst Agent for vulnerability triage. <a href="https://app.daily.dev/posts/R9WyVAVPv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft open-sources XAML Studio under .NET Foundation</strong>: Microsoft has open-sourced XAML Studio, a lightweight prototyping tool for WinUI and UWP, under the .NET Foundation, aiming for community contributions and a redesigned UI in version 2.0 with deeper WinUI 3 integration. <a href="https://app.daily.dev/posts/hUJrXK08w" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA allegedly sought pirated books for AI training</strong>: An amended authors’ lawsuit claims NVIDIA executives contacted Anna’s Archive, a pirate library, to access millions of copyrighted books for AI model training, despite warnings about the illegal nature of the content. <a href="https://app.daily.dev/posts/2IcgLruSE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI coding tools impact productivity and quality</strong>: While AI coding tools like Cursor and Copilot can accelerate app development, they often generate lower-quality code for complex tasks, requiring significant human correction and exposing ‘process debt’ in organizations lacking robust automated feedback mechanisms. <a href="https://app.daily.dev/feed-by-ids?id=3DI7vnha4&amp;id=ATZPMmQn8&amp;id=DdiXGTqPN&amp;id=L2hSWhbgY&amp;id=LJqUceZbE&amp;id=Y8gY63ibF&amp;id=aGFVTrinz&amp;id=aKVj0i6Bm&amp;id=j91dGf3TP&amp;id=tJZLXej6S" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>SaaS and DevOps downtime surges, cloud reliability questioned</strong>: SaaS downtime incidents increased by 69% in 2025, with major DevOps platforms experiencing over 9,255 hours of degraded service, highlighting critical gaps in cloud reliability and the need for robust backup and recovery strategies. <a href="https://app.daily.dev/feed-by-ids?id=HfVfar8hJ&amp;id=XatxqEJEy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft Azure Functions supports Model Context Protocol servers</strong>: Microsoft has released General Availability for Model Context Protocol (MCP) support in Azure Functions, providing secure, standardized workflows for AI agents with built-in OAuth authentication and support for multiple languages. <a href="https://app.daily.dev/posts/VULLgj3gE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>DuckDB ecosystem expands with AI agent support</strong>: The DuckDB ecosystem now supports full DML operations for Iceberg, browser-based Iceberg queries via DuckDB-Wasm, and new extensions for FIX protocol logs and graph analytics, with MotherDuck’s MCP server enabling AI agents to perform analytics through natural language. <a href="https://app.daily.dev/posts/T0eFsI0Yd" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agent psychosis: a new term for old problems</strong>: Changelog News is apparently concerned about ‘AI agent psychosis,’ a fancy new term for when AI agents act erratically. Meanwhile, <a href="http://Monday.com" target="_blank" rel="noopener nofollow">Monday.com</a> is actually engineering solutions to prevent LLM hallucinations in their Slack bot by pre-computing statistics and using human-in-the-loop approvals. It seems some folks are still busy coining dramatic new phrases while others are just, you know, building reliable systems. <a href="https://app.daily.dev/feed-by-ids?id=AJzG7x8FZ&amp;id=AS53pzdrZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Enterprise leaders love ERP, but also want to kill it</strong>: A survey says 97% of executives are satisfied with their current ERP systems, which is great. But then 70% also say traditional ERP isn’t the future, and they’re split between composable or ‘autonomous, AI-driven’ systems. So, everyone’s happy with what they have, but also wants something completely different. Sounds like a classic case of ‘we love our legacy systems until we have to pay for them,’ or maybe just a lot of buzzword bingo. <a href="https://app.daily.dev/feed-by-ids?id=QJmBgvguS&amp;id=zDjVyaxS1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Steam clarifies AI disclosure: because ‘efficiency gains’ are magic</strong>: Steam has ‘clarified’ its AI disclosure rules, now only requiring developers to fess up if AI-generated content is player-facing. AI used for ‘internal workflows, code assistance, or concept ideation’ gets a free pass. So, if AI makes your game, but the players don’t <em>see</em> the AI making it, it’s just good old human ingenuity, right? Conveniently sweeping all those ‘efficiency gains’ under the rug. <a href="https://app.daily.dev/posts/C2CcXjt7M" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
