<h2>Must know</h2>
<ul>
<li><strong>GitLab 2FA bypass and GNU telnetd root exploit</strong>: A critical vulnerability (CVE-2026-0723) in GitLab Community and Enterprise editions allows attackers to bypass two-factor authentication, leading to full account takeover. On-premises users are urged to upgrade immediately, as SaaS versions are already patched. Separately, a severe 11-year-old flaw (CVE-2026-24061) in GNU InetUtils telnetd is being actively exploited, granting remote root access. Security experts, as reported by multiple technical security outlets, recommend retiring telnetd in favor of SSH due to its inherent 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 increasingly restricting or closing their public APIs. This trend is driven by the rise of LLM data scraping and autonomous AI agents, which create monetization challenges and raise data protection concerns. The shift from open APIs to private or partner-only access signals a growing tension between open data ideals and business sustainability, potentially eroding interoperability across platforms. <a href="https://app.daily.dev/posts/Fwe6HWQAa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents are creating new insider threats</strong>: Security experts at Davos 2026 warned that AI agents pose significant insider threat risks, as they can access and manipulate sensitive data autonomously. Organizations are struggling to secure these systems and are advised to implement zero-trust architectures, continuous monitoring, and guardrails for AI agent behavior. The challenge lies in training agents to resist social engineering while maintaining their utility, with no comprehensive solutions currently available, as reported by multiple security outlets. <a href="https://app.daily.dev/posts/KOoayUI82" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI coding agents are here, but human oversight is key</strong>: AI coding agents are moving from passive tools to active collaborators, with some developers reporting 60% of their work now involves AI. However, human supervision and judgment remain essential, as only 0-20% of tasks can be fully delegated, emphasizing the need for robust architecture and strategic decisions. <a href="https://app.daily.dev/feed-by-ids?id=SxfRwPJcN&amp;id=VnW9u36zE&amp;id=SNNDpvqm6" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI scales PostgreSQL for ChatGPT demand</strong>: OpenAI successfully scaled a single primary PostgreSQL instance with nearly 50 read replicas across Azure regions to support 800 million ChatGPT users, achieving five-nines availability and low double-digit millisecond latencies by offloading write-heavy workloads to sharded CosmosDB. <a href="https://app.daily.dev/posts/3MBFUAgbd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft launches VibeVoice-ASR for long-form speech-to-text</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>New AI startup raises $480M for team coordination models</strong>: Humans&amp;, founded by ex-researchers from Anthropic, OpenAI, xAI, and Google DeepMind, raised $480 million to build foundation AI models focused on coordination and collaboration, aiming to create a ‘central nervous system’ for teams. <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 XMRig 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 after at least 1,100 downloads. <a href="https://app.daily.dev/posts/ew85R1hXm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub improves Issues performance</strong>: GitHub has significantly improved the performance of its Issues feature, with 35% of views now loading in under 200 milliseconds, up from 2% at the beginning of the year, making the experience feel ‘instant’ for signed-in users. <a href="https://app.daily.dev/posts/vT2U4VS8T" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Pulumi adds replacementTrigger for automated resource recreation</strong>: Pulumi v3.215.0 introduces <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 users. <a href="https://app.daily.dev/posts/ay0AhT4GQ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic’s Claude: now with 80 pages of feelings</strong>: Anthropic just dropped an 80-page ‘Constitution’ for Claude, up from a mere 2,700 words. Apparently, Claude is a ‘genuinely novel kind of entity’ that might have emotions and could be a ‘moral patient.’ They’re openly debating its sentience, but conveniently, there are no external expert opinions or independent assessments. Just Anthropic telling us how deep their AI’s feelings are. Right. <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: quantity over quality, as always</strong>: Cursor’s AI agents churned out a 3-million-line browser engine in a week, which sounds impressive until you hear it’s only ‘partially functional’ with an 88% CI failure rate. Industry experts, who apparently still exist, are pointing out that AI scales code production but utterly fails on quality and reliability. So, we’re trading human developers for AI that writes mountains of unreliable code. Progress! <a href="https://app.daily.dev/posts/SJiMJUgK5" 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 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 like the FTC and DOJ remain ‘silent.’ It’s almost as if the ‘collapse of content moderation standards’ is a feature, not a bug, when it comes to certain AI ventures. But hey, at least it’s fast! <a href="https://app.daily.dev/posts/EpiIORLsM" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
