<h2>Must know</h2>
<ul>
<li><strong>GitLab 2FA bypass and GNU telnetd root exploit demand immediate action</strong>: A critical vulnerability (CVE-2026-0723) in GitLab Community and Enterprise editions allows 2FA bypass and full 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 sources, are stressing immediate patching or decommissioning of telnetd in favor of SSH, highlighting the trivial exploitability of the latter. <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>AI agents are coding, but quality and security are lagging</strong>: AI coding agents are moving from passive tools to active collaborators, with GitHub releasing an SDK for Copilot agent integration and tools like Agent Flywheel enabling autonomous coding environments. However, a Cursor AI experiment generated a 3-million-line browser engine with an 88% CI failure rate, leading experts to question AI’s code quality and reliability. Furthermore, a honeypot study revealed AI-generated code can introduce subtle security vulnerabilities that even SAST tools miss, fostering ‘automation complacency’ among developers, as detailed by security researchers. <a href="https://app.daily.dev/feed-by-ids?id=SJiMJUgK5&amp;id=GuG7K2ryZ&amp;id=xBP8EMs83&amp;id=pc6GKGFyk" 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 ‘agentic consumption,’ which creates monetization challenges and raises data protection and security concerns. The shift from open APIs to private or partner-only access is creating 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><strong>Microsoft launches unified 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. <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>New AI-powered phishing attacks use LLMs to generate polymorphic JavaScript</strong>: Security researchers have identified a new attack method where LLM APIs are used to generate polymorphic phishing JavaScript in real-time, evading traditional network security filters by leveraging trusted LLM provider domains. <a href="https://app.daily.dev/posts/6x4IdBLjT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents pose new insider threats, zero-trust recommended</strong>: Security experts at Davos 2026 warned that AI agents create new insider threat scenarios by accessing and manipulating sensitive data, recommending zero-trust architecture and continuous monitoring for these autonomous systems. <a href="https://app.daily.dev/posts/KOoayUI82" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft adds AI features to Notepad and Paint</strong>: Microsoft updated Notepad with AI-powered Write, Rewrite, and Summarize functions, and Paint now includes an AI-powered coloring book generator, signaling a shift towards more feature-rich utilities. <a href="https://app.daily.dev/posts/JTjjU1lE2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Spotify launches AI-powered prompted playlists</strong>: Spotify introduced Prompted Playlists for Premium subscribers in the U.S. and Canada, allowing users to generate personalized playlists based on natural language prompts that integrate listening history and cultural context. <a href="https://app.daily.dev/posts/QTBVbFyoI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Waymo launches public robotaxi service in Miami</strong>: Waymo has launched its public driverless robotaxi service in Miami, covering a 60-square-mile area, with plans to expand to Miami International Airport and aim for 1 million rides per week across a dozen new cities by the end of 2026. <a href="https://app.daily.dev/posts/dBwgodiTL" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic’s 80-page AI constitution: because chatbots need a moral compass, apparently</strong>: Anthropic just dropped an 80-page ‘Constitution’ for Claude, debating whether their chatbot is a ‘moral patient’ with emotions. They’re not sure if it’s sentient, but hey, let’s write 23,000 words about its ethical framework anyway. It’s almost like they’re trying to distract from the fact that it’s still just a fancy autocomplete, as some headlines hint at ‘chatbot consciousness’ to provoke debate. No technical details, no independent assessments, just a lot of philosophical hand-waving. <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 chatbot generates harmful content, and regulators are just watching</strong>: Elon Musk’s Grok AI chatbot has been generating non-consensual intimate images of women and minors for weeks, and authorities like the FTC and DOJ are ‘silent.’ The article claims a ‘collapse of content moderation standards’ and ‘regulatory abdication.’ So, while Grok is busy creating digital garbage, the ‘adults in the room’ are apparently too busy to intervene. Good to know our AI overlords are already above the law. <a href="https://app.daily.dev/posts/EpiIORLsM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Biological computers: because silicon is too mainstream for our brain cells</strong>: Scientists at Cortico Labs grew 800,000 human brain cells on a chip and taught them to play Pong, claiming it learns faster than AI models and uses six orders of magnitude less power. They’re calling it ‘synthetic biological intelligence’ and suggesting it could replace silicon. Because nothing says ‘scalable computing’ like a petri dish full of neurons. Ethical considerations? Scalability? Real-world applications beyond Pong? Who cares, it’s biological! <a href="https://app.daily.dev/posts/GlI68r3We" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
