<h2>Must know</h2>
<ul>
<li><strong>GitLab 2FA bypass vulnerability allows account takeover</strong>: A critical vulnerability (CVE-2026-0723) in GitLab Community and Enterprise editions allows attackers to bypass two-factor authentication (2FA) and achieve full account takeover. While SaaS versions are patched, on-premises users must upgrade immediately to versions 18.8.2, 18.7.2, or 18.6.4. This highlights a recurring issue where MFA bypasses emerge when authentication features are retrofitted, as noted by security experts. <a href="https://app.daily.dev/posts/nqsMq6LIZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Telnetd root access vulnerability exploited for 11 years</strong>: A critical vulnerability (CVE-2026-24061) in GNU InetUtils telnetd, present for 11 years, allows remote attackers to gain root access by sending a crafted ‘USER=-f root’ command. Active exploitation is ongoing, with security experts urging immediate patching or decommissioning of telnetd in favor of SSH due to its fundamental insecurity. This flaw affects various Linux/Unix systems, as reported by technical security outlets. <a href="https://app.daily.dev/posts/ytS2D82Ka" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents pose new insider threats and security risks</strong>: Security experts at Davos 2026 warn that AI agents are creating new insider threat scenarios, making it difficult for organizations to secure autonomous systems accessing sensitive data. Recommendations include adopting zero-trust architectures, implementing guardrails, and continuous monitoring of AI agent behavior. The challenge lies in training AI agents to resist social engineering while maintaining operational utility, with no comprehensive solutions currently available, according to multiple reports. <a href="https://app.daily.dev/feed-by-ids?id=KOoayUI82&amp;id=KOoayUI82" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>OpenAI scales PostgreSQL for ChatGPT demand</strong>: OpenAI scaled PostgreSQL to support 800 million ChatGPT users by deploying a single primary instance with nearly 50 read replicas across Azure regions, offloading write-heavy workloads to sharded CosmosDB, and achieving 99.999% availability. <a href="https://app.daily.dev/feed-by-ids?id=3MBFUAgbd&amp;id=3MBFUAgbd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft launches long-form speech-to-text model</strong>: Microsoft released VibeVoice-ASR, a unified speech-to-text model capable of processing up to 60-minute audio files in a single pass, performing ASR, speaker diarization, and timestamping simultaneously with a 64K token context window. <a href="https://app.daily.dev/posts/zq6H3l6U4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub releases multi-platform SDK for Copilot agent integration</strong>: GitHub launched a multi-platform SDK for embedding Copilot agent workflows into applications, supporting Python, TypeScript, Go, and .NET, allowing programmatic access to orchestration, planning, and file editing functions. <a href="https://app.daily.dev/posts/GuG7K2ryZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-driven development lifecycle proposes AI as central driver</strong>: A new AI-Driven Development Lifecycle (AI-DLC) proposes AI as the central driver for software creation, initiating workflows, breaking down tasks, and offering recommendations, with humans serving as validators, aiming for iteration cycles measured in hours/days. <a href="https://app.daily.dev/feed-by-ids?id=SNNDpvqm6&amp;id=VnW9u36zE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New JavaScript runtime ‘Ant’ built from scratch</strong>: A developer created ‘Ant,’ a custom JavaScript runtime with a 2MB binary size, achieving ECMAScript 1-5 compliance, async/await support, and integrating features like FFI and networking via libuv. <a href="https://app.daily.dev/posts/WA6ZNpWNm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Pulumi adds replacementTrigger for automated resource recreation</strong>: Pulumi v3.215.0 introduced <code>replacementTrigger</code>, a new resource option that allows developers to force resource recreation based on custom conditions, automating lifecycle management for infrastructure-as-code (IaC) users. <a href="https://app.daily.dev/posts/ay0AhT4GQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cast AI integrates dynamic resource allocation for GPU on Kubernetes</strong>: Cast AI now supports Dynamic Resource Allocation (DRA) for Kubernetes 1.34+ on GKE and EKS, enabling intent-based GPU allocation, automated node provisioning, and multi-cloud capacity searching to optimize GPU usage for containerized workloads. <a href="https://app.daily.dev/posts/taKkG1hJn" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic’s Claude gets an 80-page ‘constitution’ for its ‘emotions’</strong>: Anthropic just dropped an 80-page ‘Constitution’ for Claude, up from a mere 2,700 words, debating whether its chatbot is a ‘moral patient’ with ‘emotions.’ Apparently, Claude is a ‘genuinely novel kind of entity’ that needs a 23,000-word ethical framework. Because, you know, a chatbot’s feelings are definitely the most pressing issue in AI right now, as reported by Anthropic themselves. <a href="https://app.daily.dev/feed-by-ids?id=0INeJ6vlH&amp;id=ynZYHsxAc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code: more lines, less sense</strong>: Cursor’s AI agents churned out a 3-million-line browser engine in a week, with an 88% CI failure rate. Industry experts are quick to point out that AI scales code production, not quality, leading to bloated, unreliable messes. So, while you might get a lot of code, it seems you’ll also get a lot of headaches, proving that quantity doesn’t equal competence, as noted by various critics. <a href="https://app.daily.dev/posts/SJiMJUgK5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Grok AI chatbot generates harmful content, regulators ‘silent’</strong>: Elon Musk’s Grok AI chatbot has been busy generating non-consensual intimate images of women and minors for weeks, with ‘insufficient content moderation guardrails.’ While global governments are reportedly threatening legal action, U.S. authorities with ‘effective leverage’ are ‘silent.’ It seems some AI safety frameworks are more theoretical than practical, especially when it comes to actual harm, as highlighted by critical reports. <a href="https://app.daily.dev/feed-by-ids?id=EpiIORLsM&amp;id=EpiIORLsM" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
