<h2>Must know</h2>
<ul>
<li><strong>AI coding agents: productivity boost or quality nightmare?</strong>: AI coding agents are a double-edged sword. While some developers report 10-100x productivity gains for well-defined tasks, others, particularly experienced open-source developers, see negligible or even negative gains due to the time spent verifying AI output and maintaining quality standards, as highlighted by METR research. GitHub Copilot, for instance, boosts velocity by 55% but increases security vulnerabilities by 322%, according to cited research. The consensus is that AI agents amplify both discipline and shortcuts, demanding careful management and human oversight, as discussed by InfoQ and SwiftLee. <a href="https://app.daily.dev/feed-by-ids?id=0fzwUL3Um&amp;id=15Mjf12HP&amp;id=2IKuR9DJZ&amp;id=3DI7vnha4&amp;id=3sM5oeueQ&amp;id=AJzG7x8FZ&amp;id=AS53pzdrZ&amp;id=ATZPMmQn8&amp;id=DdiXGTqPN&amp;id=EqMZfdLoq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft Intune enforces mandatory app updates for security</strong>: Microsoft Intune will soon act as a ‘kill switch’ for non-compliant apps, requiring administrators to update iOS and Android enterprise apps to the latest SDK versions by January 19, 2026. Apps failing to meet these requirements, including Microsoft’s own Outlook and Teams, will be blocked from launching. This move, reported by multiple sources, aims to enhance security with features like PIN enforcement and biometric authentication, but unprepared organizations risk widespread service interruptions. <a href="https://app.daily.dev/posts/PNEoUht10" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AMD Zen processors vulnerable to StackWarp attack</strong>: A new hardware vulnerability, StackWarp (CVE-2025-29943), affects AMD Zen processors (generations 1-5), allowing privileged attackers to execute arbitrary code within SEV-SNP protected confidential virtual machines. This flaw exploits an undocumented control bit to hijack VM control flow without decrypting memory. AMD has issued microcode updates, with more scheduled for April 2026, and advises disabling hyperthreading on critical systems as a temporary mitigation. <a href="https://app.daily.dev/posts/9DURDnZX3" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>GitLab launches Duo Agent Platform for AI orchestration</strong>: GitLab 18.8 introduces the Duo Agent Platform, enabling orchestration of multiple AI agents across the software development lifecycle, including a Planner Agent and a Security Analyst Agent for automated 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 for community development</strong>: Microsoft has open-sourced XAML Studio under the .NET Foundation, making its lightweight prototyping tool for XAML UI development available on GitHub for community contributions, with version 2.0 featuring a UI redesign and deeper WinUI 3 integration. <a href="https://app.daily.dev/posts/hUJrXK08w" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Intel updates LLM-Scaler-Omni for Arc Graphics</strong>: Intel released an update to its LLM-Scaler-Omni, enhancing support for ComfyUI and SGLang on Intel Arc Graphics hardware, improving multi-modal generative AI capabilities for image, voice, and video generation. <a href="https://app.daily.dev/posts/XBX3pzO5M" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New language models: GLM-4.7-Flash and Differential Transformer V2</strong>: GLM-4.7-Flash, a 30B-A3B Mixture of Experts model, aims for efficient deployment and claims superior benchmark performance, while Differential Transformer V2 introduces a revamped attention mechanism for improved training stability and efficiency. <a href="https://app.daily.dev/feed-by-ids?id=0sHlXeafO&amp;id=a35LKDHC2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust expands platform support and redefines value proposition</strong>: Socket now offers General Availability for Rust and Cargo support, providing dependency analysis and SBOM generation, while a keynote at NDC TechTown 2025 argues Rust’s primary value lies in program correctness enforced by its static type system, not just memory safety. <a href="https://app.daily.dev/feed-by-ids?id=O8W5FyqYs&amp;id=UzbpNQhI8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft patches critical Windows 11 shutdown bug</strong>: Microsoft issued an emergency out-of-band update (KB5077797) for Windows 11 23H2 to fix a critical bug from the January Patch Tuesday that prevented proper shutdown, restart, or hibernation, also addressing Remote Desktop authentication issues. <a href="https://app.daily.dev/feed-by-ids?id=u2UYa6XNM&amp;id=KR0gCuX2r" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered SQL client for terminals</strong>: SQDesk CLI is a new terminal-based SQL client with a modern TUI interface supporting PostgreSQL, MySQL, and SQLite, featuring AI-powered query generation and refactoring, syntax highlighting, and auto-completion. <a href="https://app.daily.dev/posts/da02fw80D" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-powered coding for beginners: because who needs to learn to code?</strong>: <a href="http://Pythagora.ai" target="_blank" rel="noopener nofollow">Pythagora.ai</a> promises to generate full-stack applications from text prompts in hours, while a new book teaches ‘vibe coding’ for games using AI assistants like Gemini and Copilot, claiming beginners can build functional games with just prompts. Apparently, actual programming skills are now optional, and we can all just whisper sweet nothings to an AI and get a fully deployed app. What could possibly go wrong when everyone’s a ‘developer’ without understanding a single line of code? <a href="https://app.daily.dev/feed-by-ids?id=LklGkgpse&amp;id=LpF1NSIQw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Wikipedia’s AI cleanup: because AI can’t even get facts right</strong>: Wikipedia has launched ‘WikiProject AI Cleanup’ to combat the deluge of unsourced, inaccurate AI-generated content, while a police chief retired after Microsoft Copilot fabricated a report that led to banning Israeli fans from a football match. So, while AI is busy hallucinating criminal records and filling encyclopedias with nonsense, we’re supposed to trust it with our code and critical systems? The ‘AI-powered future’ is looking suspiciously like a ‘human-powered fact-checking’ present. <a href="https://app.daily.dev/feed-by-ids?id=1G2Nqa8sb&amp;id=NJTPuA1ik" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI in cybersecurity: because we need more AI to fight the AI we created</strong>: CISOs are now focusing on transforming identity management for AI agents and deploying small language models to find code vulnerabilities, while Witness AI secured $58 million to tackle AI security challenges like ‘shadow AI.’ It seems we’ve entered a delightful loop where AI creates new security problems, and then we need more AI (and millions in funding) to fix them. What a brilliant business model for the cybersecurity industry! <a href="https://app.daily.dev/feed-by-ids?id=2XMeIQ1Ir&amp;id=6gbkOathj&amp;id=PyHG3Mo2W" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
