<h2>Must know</h2>
<ul>
<li><strong>Ransomware gangs weaponize compliance violations</strong>: Ransomware groups like Anubis and RansomHub are now threatening to report victims’ compliance violations to regulators, especially in highly regulated sectors like healthcare. This tactic, detailed in multiple reports, creates a double bind: pay the ransom or face regulatory penalties and reputational damage. Attackers are using AI to analyze stolen data for compliance gaps and even draft legal complaints, making this a rapidly adopted and effective new extortion method. <a href="https://app.daily.dev/posts/OsQyB9bxn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cisco patches critical zero-day RCE 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 since late November 2025 by the China-linked APT group UAT-9686, allows arbitrary command execution as root if the Spam Quarantine feature is internet-exposed. Cisco’s fix also aims to remove persistence mechanisms installed by attackers, as reported by security outlets. <a href="https://app.daily.dev/posts/Faoj4Wg3E" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS CodeBuild webhook 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 SDK repositories, potentially leading to global supply chain attacks. AWS patched the vulnerability within 48 hours of notification and found no evidence of in-the-wild exploitation, according to Wiz and AWS statements. <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>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 TSMC’s largest client. <a href="https://app.daily.dev/posts/aTc9ls9yL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI invests in Sam Altman’s brain-computer interface startup</strong>: OpenAI led 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>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 via Wikimedia Enterprise, monetizing structured access to Wikipedia data for training large language models. <a href="https://app.daily.dev/posts/QmEAqVcYI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub Copilot introduces agentic memory in Visual Studio</strong>: GitHub Copilot now features ‘memories’ that learn and document coding preferences and project guidelines, aiming to reduce code review friction and streamline onboarding by sharing learned patterns across AI agents. <a href="https://app.daily.dev/posts/nCmOnsQS3" 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 on Raspberry Pi 5, emphasizing privacy and low-latency. <a href="https://app.daily.dev/posts/nbjHwDlw3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Oracle faces class action over undisclosed debt for OpenAI datacenters</strong>: Oracle is being sued by bondholders who allege the company failed to disclose an additional $38 billion in debt intended for datacenter investments supporting a $300 billion deal with OpenAI, leading to bond downgrades. <a href="https://app.daily.dev/posts/9fgRHJtn9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Taiwan to invest $250 billion in US semiconductor and AI</strong>: Taiwan will invest $250 billion in US semiconductor, energy, and AI production, along with $250 billion in credit guarantees, as part of a trade deal aimed at reducing US reliance on foreign semiconductor 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>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-powered DevOps: because manual errors weren’t exciting enough</strong>: Apparently, we’re now integrating AI into DevOps, promising to automate everything from Linux to Kubernetes and Terraform. Telusko is even offering a four-month course on ‘AI-Powered DevOps,’ implying that the existing complexity wasn’t enough of a challenge. I’m sure adding probabilistic AI to our infrastructure management will only lead to more predictable outcomes, not entirely new categories of debugging nightmares. <a href="https://app.daily.dev/posts/SptJXtiZK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Agentic AI and non-human identity security: more buzzwords, same old problems</strong>: We’re told that ‘agentic AI systems’ are the future for managing ‘non-human identities’ and their secrets, using machine learning for anomaly detection. Yet, organizations are simultaneously facing a lack of visibility into ‘shadow agents’ and a ‘risk sandwich’ where scaling outpaces governance. So, we’re building more complex AI systems to secure other AI systems, while admitting we can’t even track the ones we already have. Sounds like a perfectly secure and manageable plan. <a href="https://app.daily.dev/feed-by-ids?id=NCcbuPG8X&amp;id=NVsuaRFG5&amp;id=UAGIEGe4I" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Sensitive corporate data exposed via GenAI: who could have seen this coming?</strong>: Harmonic Security analyzed millions of prompts and ‘shockingly’ found that 2.6% 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. It’s almost as if feeding proprietary information into black-box AI services without proper controls was a bad idea. But hey, at least we have a new security vendor to tell us what we already knew. <a href="https://app.daily.dev/feed-by-ids?id=F9SzKFotn&amp;id=mZZUloj80" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
