<h2>Must know</h2>
<ul>
<li><strong>GitHub Copilot gets memory, learns your coding habits</strong>: GitHub Copilot is rolling out a new ‘memories’ feature in Visual Studio, allowing it to learn and document coding preferences and project guidelines automatically. This aims to reduce code review friction and streamline onboarding by maintaining consistency, as detailed in the Visual Studio Blog. The system attaches code citations and verifies them in real-time, ensuring recommendations are always up-to-date. It’s now in public preview for all paid plans, with options for individual and organizational control. <a href="https://app.daily.dev/feed-by-ids?id=nCmOnsQS3&amp;id=s6qcADuaI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ransomware gangs weaponize compliance violations</strong>: Ransomware groups like Anubis and RansomHub are now threatening victims, especially in highly regulated sectors like healthcare, with regulatory reporting of compliance violations. This new tactic creates a dual threat: ransom demands and the risk of regulatory punishment for compliance failures. Attackers are using AI to analyze stolen data for compliance gaps and even generate formal legal complaints, according to multiple reports. This significantly increases reputational and financial risk for victim organizations. <a href="https://app.daily.dev/posts/OsQyB9bxn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cisco patches critical zero-day exploited by China-linked APT</strong>: Cisco has released emergency patches for a critical remote command execution (RCE) vulnerability (CVE-2025-20393) in AsyncOS for Secure Email Gateway and Secure Email and Web Manager. This flaw, actively exploited by the China-linked APT group UAT-9686 since late November 2025, allows arbitrary command execution as root if the Spam Quarantine feature is internet-exposed. Cisco’s fix also removes persistence mechanisms installed by attackers, as reported by various security outlets. <a href="https://app.daily.dev/posts/Faoj4Wg3E" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI-powered security platform finds zero-days at speed</strong>: Trend Micro’s TrendAI has launched ÆSIR, an AI-powered security research platform that has already discovered 21 critical CVEs across NVIDIA, Tencent, and MLflow since mid-2025, combining automated vulnerability discovery with human oversight. <a href="https://app.daily.dev/posts/Rj9ax7flK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS CodeBuild vulnerability exposed global cloud supply chain</strong>: Wiz researchers discovered ‘CodeBreach,’ a critical flaw in AWS CodeBuild’s webhook regex filters that could have led to unauthorized access and supply chain attacks on AWS GitHub repositories, though AWS patched it within 48 hours with no evidence of in-the-wild exploitation. <a href="https://app.daily.dev/feed-by-ids?id=q7gLeN4EH&amp;id=MZnlHztN0&amp;id=PTMQTWkW2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI invests in Altman’s brain-computer interface startup</strong>: OpenAI participated in a $250 million seed round for Merge Labs, a non-invasive brain-computer interface (BCI) startup co-founded by Sam Altman, raising potential conflict of interest concerns given Altman’s dual roles. <a href="https://app.daily.dev/posts/rYkFwNLz3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>TSMC revenue surges on AI demand, Nvidia becomes top client</strong>: TSMC reported 25.5% year-over-year revenue growth in Q4 2025, forecasting 30% growth for FY 2026, driven by persistent AI infrastructure demand, with Nvidia reportedly surpassing Apple as its largest client. <a href="https://app.daily.dev/posts/aTc9ls9yL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Raspberry Pi releases AI HAT+ 2 for on-device generative AI</strong>: Raspberry Pi launched the AI HAT+ 2 add-on board with a Hailo-10H neural network accelerator, enabling local execution of smaller LLMs and computer vision models with 8GB onboard RAM, targeting hobbyists and edge computing. <a href="https://app.daily.dev/posts/nbjHwDlw3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Oracle sued over undisclosed debt for OpenAI datacenter investments</strong>: Oracle faces a class action lawsuit from bondholders alleging the company failed to disclose an additional $38 billion in debt intended for datacenter investments supporting a $300 billion deal with OpenAI, causing bond value declines. <a href="https://app.daily.dev/posts/9fgRHJtn9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Wikimedia partners with major AI companies for data access</strong>: The Wikimedia Foundation has launched commercial partnerships with Amazon, Meta, Microsoft, Mistral AI, and Perplexity through Wikimedia Enterprise, monetizing structured access to Wikipedia content for training large language models. <a href="https://app.daily.dev/posts/QmEAqVcYI" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agents: The new ‘coworkers’ who never complain (or get paid)</strong>: Apparently, the future of coding is messaging an AI like a coworker to fix bugs and create pull requests, all from Slack. Kilo’s new integration, supporting models like Opus 4.5, promises to eliminate IDEs and make AI a ‘transparent team member.’ Because nothing says ‘efficient workflow’ like delegating critical engineering tasks to a chatbot that ‘struggles with complex architectures and long-running reasoning tasks,’ as MetalBear’s engineers found. What could possibly go wrong with treating an LLM as a peer? <a href="https://app.daily.dev/feed-by-ids?id=AM0JrslAM&amp;id=cFlscGQW4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered DevOps: Because humans are too slow for the cloud</strong>: Telusko is offering a four-month live course on ‘AI-Powered DevOps,’ promising to integrate AI for productivity across Linux, AWS, Docker, Kubernetes, and more. Because clearly, the problem with DevOps wasn’t complexity or human error, but a lack of AI. Never mind that AI security risks are shifting from models to ‘workflow security,’ as The Hacker News points out, with attacks manipulating context and integrations. Let’s just throw more AI at it and hope for the best! <a href="https://app.daily.dev/feed-by-ids?id=SptJXtiZK&amp;id=mZZUloj80" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GNOME 50 Alpha guts X11 code, but the timeline is a bit… off</strong>: GNOME 50 Alpha has boldly removed all X11 display server code, signaling a clear direction for the Linux desktop environment. However, the articles state the Alpha was released in January 2026, while the official stable release is scheduled for March 2025. Either GNOME has mastered time travel, or someone needs to check their calendar. Details on alternatives like Wayland are conspicuously absent, leaving us to wonder if this is a bold leap forward or just a premature removal. <a href="https://app.daily.dev/posts/SmvPruPWQ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
