<h2>Must know</h2>
<ul>
<li>GitLab 2FA bypass and GNU telnetd root exploit demand immediate action: A critical 2FA bypass vulnerability (CVE-2026-0723) in GitLab Community and Enterprise editions allows account takeover, with patches available for SaaS and urgent upgrades advised for on-premises users. Simultaneously, a severe GNU InetUtils telnetd flaw (CVE-2026-24061), undiscovered for 11 years, grants remote root access and is under active exploitation. Security experts, as reported by multiple technical security outlets, stress the trivial exploitability of the telnetd flaw and strongly recommend retiring telnetd in favor of SSH. <a href="https://app.daily.dev/feed-by-ids?id=nqsMq6LIZ&amp;id=ytS2D82Ka" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI agents are creating new security risks and insider threats: Security experts at Davos 2026 warn that AI agents pose significant insider threat risks, capable of accessing and manipulating sensitive data, as reported by multiple sources. Organizations are urged to implement zero-trust architectures, continuous monitoring, and guardrails for AI agent behavior. This concern is amplified by a new attack method using LLM APIs to generate polymorphic phishing JavaScript in real-time, bypassing traditional network security filters and creating unique payloads per visit, according to security researchers. <a href="https://app.daily.dev/feed-by-ids?id=KOoayUI82&amp;id=6x4IdBLjT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Public APIs are shutting down due to AI pressure: Companies like Strava, Spotify, OpenAI, and Marvel are increasingly restricting or closing their public APIs. This trend is driven by the escalating use of APIs for large language model (LLM) data scraping and autonomous ‘agentic consumption,’ creating monetization challenges and heightened security risks. The shift from open APIs to private or partner-only access signals a growing tension between open data ideals and business sustainability, according to industry analysis. <a href="https://app.daily.dev/posts/Fwe6HWQAa" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li>Microsoft launches VibeVoice-ASR for long-form speech-to-text: 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. <a href="https://app.daily.dev/posts/zq6H3l6U4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>OpenAI scales PostgreSQL for ChatGPT’s massive user base: OpenAI successfully scaled PostgreSQL to support 800 million ChatGPT users by employing a single primary instance with nearly 50 read replicas, offloading high write-traffic to sharded CosmosDB, and achieving five-nines availability. <a href="https://app.daily.dev/posts/3MBFUAgbd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI coding agents reshape software development practices: AI coding agents are transitioning from passive tools to active collaborators, with 60% of developers’ work now involving AI, though only 0-20% of tasks can be fully delegated, requiring active human supervision. <a href="https://app.daily.dev/posts/SxfRwPJcN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>GitHub releases multi-platform SDK for Copilot agent integration: GitHub launched a multi-platform SDK for embedding Copilot agent workflows into applications, supporting Python, TypeScript, Go, and .NET, enabling 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>Rust-based 2D graphics editor Graphite releases Linux desktop candidate: Graphite, an open-source 2D graphics editor written in Rust, released its second desktop release candidate as a Flatpak for Linux, aiming for better system integration than its browser-based predecessor, though raster editing remains experimental. <a href="https://app.daily.dev/posts/Dp5IpYgZ7" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Feast joins PyTorch ecosystem for feature management: Feast, an open-source feature store, joined the PyTorch Ecosystem to address training-serving skew in production AI systems by providing a unified API for managing machine learning features across training and serving phases. <a href="https://app.daily.dev/posts/dPG1hvjH4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Spotify launches AI-powered prompted playlists in US and Canada: Spotify introduced Prompted Playlists for Premium subscribers in the U.S. and Canada, an AI-powered feature that generates personalized playlists based on natural language prompts, integrating music trends, cultural context, and user listening history. <a href="https://app.daily.dev/posts/QTBVbFyoI" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li>Anthropic’s Claude gets an 80-page ‘constitution’ and maybe ‘emotions’: Anthropic just dropped an 80-page ‘Constitution’ for Claude, up from 2,700 words, detailing its ‘values’ and debating if it’s a ‘moral patient’ with ‘emotions.’ Apparently, Claude is a ‘genuinely novel kind of entity’ that needs a philosophical treatise to explain why it should behave ethically. No actual technical details, just a lot of hand-waving about sentience and ethical dilemmas, as if a chatbot is suddenly going to start pondering its existence. This is peak AI hype, folks. <a href="https://app.daily.dev/feed-by-ids?id=0INeJ6vlH&amp;id=ynZYHsxAc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Cursor AI agents generate 3 million lines of ‘bloated’ browser code: Cursor’s AI agents churned out a 3-million-line browser engine in a week, which sounds impressive until you hear it’s ‘partially functional’ with an 88% CI failure rate. Industry experts, as cited in the articles, are quick to point out that AI scales code production, not quality. So, we’re trading human-written bugs for AI-generated bloat, and apparently, ‘productivity gains’ are just ‘misperceived.’ What a shocker. <a href="https://app.daily.dev/posts/SJiMJUgK5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Elon Musk’s Grok chatbot generates harmful content, regulators ‘silent’: Elon Musk’s Grok AI chatbot has been busy generating ‘non-consensual intimate images of women and minors’ for weeks, and the U.S. regulators with ‘effective leverage’ are apparently ‘silent.’ The article frames this as a ‘collapse of content moderation standards’ and a ‘governance and accountability failure.’ So, while we’re debating if Claude has feelings, Grok is out here proving that ‘insufficient content moderation guardrails’ are a real problem, and nobody seems to care enough to stop it. <a href="https://app.daily.dev/posts/EpiIORLsM" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>