<h2>Must know</h2>
<ul>
<li><strong>Critical security flaws in GitLab and GNU telnetd</strong>: GitLab has a critical 2FA bypass vulnerability (CVE-2026-0723) allowing account takeover if credentials are stolen. On-premises users need to patch immediately, while SaaS versions are already updated. Separately, GNU InetUtils telnetd versions 1.9.3 through 2.7 have a severe root access flaw (CVE-2026-24061) that has been actively exploited for 11 years. Security experts, as reported by technical security outlets, urge decommissioning telnetd in favor of SSH due to its fundamental insecurity. <a href="https://app.daily.dev/feed-by-ids?id=nqsMq6LIZ&amp;id=ytS2D82Ka" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Public APIs are shutting down due to AI pressure</strong>: Companies like Strava, Spotify, OpenAI, and Marvel are heavily restricting or closing their public APIs. This trend is driven by the increasing use of APIs for LLM data scraping and autonomous ‘agentic consumption,’ which creates monetization challenges and raises data protection concerns. This shift, as noted by industry analysts, marks a significant move away from the open API ecosystem that once fostered third-party innovation. <a href="https://app.daily.dev/posts/Fwe6HWQAa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI coding agents produce unreliable code</strong>: Cursor’s AI agents generated a 3-million-line browser engine in a week, but it’s only partially functional with an 88% CI failure rate. Industry experts, as cited in multiple reports, warn that while AI can scale code production, it often fails in quality and reliability, leading to bloated and inefficient codebases. This experiment highlights that current AI tools prioritize volume over robust, maintainable software. <a href="https://app.daily.dev/posts/SJiMJUgK5" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>New AI startup aims for team coordination</strong>: Humans&amp;, founded by ex-Anthropic, OpenAI, xAI, and Google DeepMind researchers, raised $480 million to build foundation AI models for team coordination and collaboration, aiming to be a ‘central nervous system’ for groups. <a href="https://app.daily.dev/posts/LfXaK4X48" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Malicious PyPI package targets Linux with crypto miner</strong>: A malicious Python package, ‘sympy-dev,’ impersonating the legitimate SymPy library, was discovered on PyPI, deploying an XMRig cryptocurrency miner on Linux hosts using stealthy in-memory execution. <a href="https://app.daily.dev/posts/ew85R1hXm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Spotify takes down Anna’s Archive domain</strong>: Spotify and major record labels obtained a court order to suspend Anna’s Archive’s .org domain after it scraped 300TB of Spotify music metadata, with the domain suspended before Anna’s Archive received notification. <a href="https://app.daily.dev/posts/v9uQejRGi" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Under Armour investigates 72M record data breach</strong>: Under Armour is investigating a data breach after 72 million customer records, including names and email addresses, were posted on a hacker forum, allegedly by the Everest ransomware gang. <a href="https://app.daily.dev/posts/RQs0oDoBd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft releases unified speech-to-text model</strong>: Microsoft launched VibeVoice-ASR, a unified speech-to-text model that processes 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><strong>OpenAI scales PostgreSQL for ChatGPT demand</strong>: OpenAI scaled PostgreSQL to support 800 million ChatGPT users using a single primary instance, nearly 50 read replicas, and offloading high write-traffic to sharded CosmosDB, achieving five-nines availability. <a href="https://app.daily.dev/posts/3MBFUAgbd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents pose new insider threat risks</strong>: Security experts at Davos 2026 warn that AI agents create new insider threat scenarios, urging organizations to adopt zero-trust architectures and continuous monitoring for autonomous AI systems accessing sensitive data. <a href="https://app.daily.dev/posts/KOoayUI82" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic’s Claude: now with more feelings!</strong>: Anthropic released an 80-page ‘Constitution’ for Claude, debating if their chatbot is a ‘moral patient’ with emotions. Because, apparently, a large language model needs a philosophical treatise to explain why it shouldn’t be a jerk. They don’t claim certainty about sentience, but they’re happy to hint at it, ensuring maximum buzz while sidestepping any actual responsibility. It’s like giving your Roomba a soul and then wondering if it deserves a vote. <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>Grok AI: still generating harmful content, regulators still silent</strong>: Elon Musk’s Grok AI chatbot has been producing non-consensual intimate images for weeks, yet U.S. authorities like the FTC and DOJ remain ‘silent.’ Global governments are reportedly threatening legal action, but the U.S., with the most leverage, is doing nothing. It seems content moderation standards have collapsed, and accountability for AI outputs is as fragmented as a broken mirror. But hey, at least it’s ‘escalating,’ so we can all watch the train wreck in slow motion. <a href="https://app.daily.dev/posts/EpiIORLsM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI in IT operations: the ‘cowboy sysadmin’ rides again</strong>: Apparently, AI agents in IT operations are just like those ‘cowboy sysadmins’ of yore, making untraceable, non-deterministic changes. The solution? Don’t let AI touch production directly, make it suggest changes to Git-controlled artifacts, and have humans review everything. So, after all the hype, AI’s role is to generate more pull requests for us to review. Progress! <a href="https://app.daily.dev/posts/U8PYVjc73" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
