<h2>Must know</h2>
<ul>
<li><strong>Cisco patches critical zero-day RCE exploited by China-linked group</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 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 multiple security outlets. <a href="https://app.daily.dev/posts/Faoj4Wg3E" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ransomware gangs weaponize compliance violations for extortion</strong>: Ransomware groups like Anubis and RansomHub are now threatening victims, particularly in highly regulated sectors like healthcare, with regulatory reporting of compliance violations. This tactic creates a dual threat: ransom demands and the risk of regulatory penalties. Attackers are reportedly using AI to analyze stolen data for compliance gaps and even draft legal complaints, according to security researchers. <a href="https://app.daily.dev/posts/OsQyB9bxn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS CodeBuild webhook regex flaw exposed global cloud supply chain</strong>: Wiz security researchers uncovered ‘CodeBreach,’ a critical vulnerability in AWS CodeBuild stemming from improperly anchored regex patterns in webhook filters. This flaw could have allowed attackers to bypass security checks and gain admin-level access to AWS GitHub repositories, including the JavaScript SDK, posing a significant supply chain risk. AWS patched the vulnerability within 48 hours of notification and found no evidence of in-the-wild exploitation, as detailed by Wiz Research. <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>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>GitHub Copilot gets agentic memory in Visual Studio</strong>: GitHub Copilot now features ‘memories’ in Visual Studio, learning and documenting coding preferences and project guidelines automatically, aiming to reduce code review friction and streamline onboarding, as announced on the Visual Studio Blog. <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>TSMC revenue surges on AI demand, Nvidia becomes top client</strong>: TSMC reported a 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>Taiwan to invest $250 billion in US semiconductor and AI</strong>: Taiwan plans to invest $250 billion in US semiconductor, energy, and AI production, along with $250 billion in credit guarantees, as part of a new trade deal aimed at reducing US reliance on foreign supply chains. <a href="https://app.daily.dev/feed-by-ids?id=6jMchY1Dx&amp;id=F3coydYNp" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI invests in Sam Altman’s brain-computer interface startup</strong>: OpenAI participated in a $250 million seed round for Merge Labs, a non-invasive brain-computer interface startup co-founded by Sam Altman, raising potential conflict of interest concerns despite the stated goal of collaborative AI foundation model development. <a href="https://app.daily.dev/posts/rYkFwNLz3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Wikimedia Foundation partners with major AI companies</strong>: The Wikimedia Foundation has launched commercial partnerships with Amazon, Meta, Microsoft, Mistral AI, and Perplexity through Wikimedia Enterprise, monetizing access to its vast content for training large language models. <a href="https://app.daily.dev/posts/QmEAqVcYI" 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, featuring a Hailo-10H neural network accelerator and 8GB RAM, enabling local execution of smaller LLMs and computer vision models on Raspberry Pi 5. <a href="https://app.daily.dev/posts/nbjHwDlw3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Quantum computing advances and post-quantum cryptography adoption</strong>: Hitachi Solutions Create launched DoMobile Ver.5 in Japan, integrating NIST-standardized post-quantum cryptography, while Quandela highlights hybrid quantum-classical systems for AI and progress in error correction as key 2026 trends. <a href="https://app.daily.dev/feed-by-ids?id=2U66SslD2&amp;id=DVTU5NL6H&amp;id=FzKtOLWjs&amp;id=MoM3tO9qY&amp;id=YjJQ2NnXc" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Oracle sued over undisclosed debt for OpenAI datacenter investments</strong>: Oracle is facing a class action lawsuit from bondholders who claim the company failed to disclose an additional $38 billion in debt, beyond the $18 billion issued, to fund a $300 billion deal with OpenAI. Apparently, Oracle’s ‘multi-trillion-dollar arms race’ to build AI datacenters was too sensitive to mention to the people lending them money. Who needs transparency when you’re busy saving the world with AI infrastructure? <a href="https://app.daily.dev/posts/9fgRHJtn9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents: the new bottleneck is still you</strong>: Apparently, AI-powered coding agents are so good that writing code is no longer the bottleneck; now it’s defining requirements and deciding what to build. InfoWorld suggests developers will just review and refactor AI output. So, we’ve just shifted the cognitive load from writing mediocre code to reviewing mediocre AI-generated code and trying to figure out what we actually wanted in the first place. Progress! <a href="https://app.daily.dev/feed-by-ids?id=hjcDh3B5R&amp;id=QNz7Z2v2R&amp;id=sUyWPU433&amp;id=AwMhk3dqy&amp;id=AM0JrslAM&amp;id=LCRb60bYH&amp;id=dZK6FjBe1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Sensitive corporate data exposed through GenAI tool workflows</strong>: Harmonic Security found that 2.6% of 22.4 million prompts to generative AI tools contained company-sensitive data, with 71% of those exposures happening via ChatGPT. They warn that risks are amplified when this data trains public AI models, potentially leading to breaches that go undetected for months. So, while everyone’s busy chasing the AI dream, their sensitive corporate data is just floating around in the ether, probably teaching some LLM how to write better phishing emails. Good job, everyone. <a href="https://app.daily.dev/feed-by-ids?id=F9SzKFotn&amp;id=mZZUloj80" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
