<h2>Must know</h2>
<ul>
<li><strong>Ransomware gangs weaponize compliance violations</strong>: Ransomware groups like Anubis and RansomHub are now threatening victims, especially in healthcare, with regulatory reporting of compliance violations if ransoms aren’t paid. This tactic, which leverages AI to analyze stolen data for compliance gaps and draft legal complaints, creates a dual threat of financial and regulatory penalties. Organizations are forced to choose between paying up or facing public exposure of their compliance failures, as reported across multiple sources. <a href="https://app.daily.dev/posts/OsQyB9bxn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI invests in 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 OpenAI CEO Sam Altman. Merge Labs aims to develop BCIs using molecular neuron interfacing and ultrasound, positioning itself against invasive methods like Neuralink. While OpenAI sees this as a synergistic investment for AI foundation model development, it raises significant governance and ethics concerns due to Altman’s dual role, especially given OpenAI’s projected $74 billion operating loss by 2028, according to reports. <a href="https://app.daily.dev/posts/rYkFwNLz3" 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 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 recommends immediate patching and highlights that exploited systems may have persistence mechanisms installed by attackers. <a href="https://app.daily.dev/posts/Faoj2Wg3E" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AWS CodeBuild webhook flaw exposed global cloud supply chain</strong>: Wiz researchers discovered ‘CodeBreach,’ a critical vulnerability in AWS CodeBuild due to improperly anchored regex patterns, which could have allowed attackers to gain admin-level access to AWS SDK repositories and potentially launch global supply chain attacks. AWS patched the flaw within 48 hours and found 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>GitHub Copilot gets 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 providing context-aware recommendations and verifying them against the active codebase. <a href="https://app.daily.dev/posts/nCmOnsQS3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Raspberry Pi AI HAT+ 2 enables on-device generative AI</strong>: Raspberry Pi released the AI HAT+ 2 add-on board, equipped with a Hailo-10H neural network accelerator and 8GB RAM, enabling local execution of smaller LLMs and computer vision models on Raspberry Pi 5, eliminating cloud dependency for edge AI applications. <a href="https://app.daily.dev/posts/nbjHwDlw3" 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 via Wikimedia Enterprise, monetizing structured access to Wikipedia data for training large language models and addressing rising operational costs. <a href="https://app.daily.dev/posts/QmEAqVcYI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Oracle faces class action over undisclosed debt for AI datacenters</strong>: Oracle is being sued by bondholders who allege the company failed to disclose an additional $38 billion in debt for datacenter investments tied to a $300 billion deal with OpenAI, leading to bond downgrades and investor losses. <a href="https://app.daily.dev/posts/9fgRHJtn9" 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>US imposes 25% tariff on AI semiconductors to China</strong>: The Trump administration has imposed a 25% tariff on advanced AI semiconductors, including Nvidia’s H200 and AMD’s MI325X, produced outside the US but routed through the US for export to China, aiming to reduce reliance on foreign supply chains. <a href="https://app.daily.dev/posts/BdyAfYrKk" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agents: the new security disaster in the making</strong>: Apparently, the rush to deploy AI agents is creating a security nightmare. We’re seeing critical vulnerabilities like ServiceNow’s ‘BodySnatcher’ allowing unauthenticated admin impersonation and weaponized AI agents. It seems everyone’s so busy chasing the AI hype train that basic security, like zero-trust for AI agents and proper lifecycle management, is an afterthought. Who needs secure systems when you can just ship fast and break things, right? As Security Boulevard points out, this is a disaster waiting to happen. <a href="https://app.daily.dev/posts/M77gEo1eY" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI coding tools: the great productivity myth?</strong>: A Stanford researcher found AI coding tools only boost productivity by 10-15% on average, with some teams seeing minimal or even negative impact. And get this: ‘ghost engineers’—full-time developers doing next to nothing—make up 9% of teams, more if they’re remote. So, while AI can churn out code, it seems the real problem isn’t writing it, but having actual humans who know what they’re doing. Maybe we should focus on that instead of chasing marginal AI gains, as one researcher suggests. <a href="https://app.daily.dev/posts/jSgoQLSoF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microservices: solving organizational problems, creating technical ones</strong>: Turns out, those shiny microservices everyone’s been gushing about are actually killing performance. Network calls add 1,000-5,000x latency, leading to 50-150% higher latency, 300% more resource usage, and double the infrastructure costs. It seems microservices are more about making managers happy with ‘team autonomy’ than actually building efficient systems. So, unless you’re a massive enterprise with 50+ engineers, maybe stick to a modular monolith and save yourself the headache, as one article bluntly puts it. <a href="https://app.daily.dev/posts/OC7oxM66D" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
