<h2>Must know</h2>
<ul>
<li><strong>AI agent security is a dumpster fire, not a new problem</strong>: The experimental AI social network Moltbook collapsed due to catastrophic security flaws like open database access, exposing millions of API tokens and emails. Security experts from Wiz and Snyk highlight that these aren’t novel AI problems but traditional security failures: exposed databases, lack of guardrails, and over-privileged access. This incident, along with the discovery of 36% of AI agent skills on ClawHub having security flaws, underscores that the rush to deploy AI agents is bypassing fundamental security practices, creating familiar vulnerabilities at an accelerated pace. Meanwhile, GitHub’s Agent HQ and OpenAI’s Codex app are pushing multi-agent orchestration, but without addressing these foundational security issues, we’re just building bigger, faster-collapsing sandcastles. <a href="https://app.daily.dev/feed-by-ids?id=ycPDVy6OP&amp;id=yCqSQa4jM&amp;id=8XBbHzIL3&amp;id=8XvGhTS0d&amp;id=BFb4A0CqB&amp;id=OTgLY8MMT&amp;id=Oalw1HqbC&amp;id=19cH7clY0&amp;id=0BdgmYNfg&amp;id=0UIAIJz1z" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI is eating the SaaS world, and traditional UIs are next</strong>: Anthropic’s Claude Co-work plugins triggered a $400 billion stock drop in software companies, demonstrating how AI agents can replace traditional SaaS UIs by handling business logic at the AI tier. Microsoft’s CEO predicted this shift, where AI agents orchestrate CRUD databases, making many feature-focused SaaS products vulnerable. Companies like Databricks are already pivoting to AI-native solutions, arguing that AI will make traditional UIs irrelevant, shifting value from tool access to outcome-based models. This isn’t just about ads in ChatGPT; it’s a fundamental re-architecture of how software delivers value, forcing SaaS companies to become systems of record or platforms for customization to survive. <a href="https://app.daily.dev/feed-by-ids?id=QnkkrfiEd&amp;id=zMRAHcD2R&amp;id=1TELuPW5m&amp;id=5kN0YzGRy&amp;id=DArhVFauV&amp;id=oAAn2Vx2M&amp;id=oAAn2Vx2M" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux kernel 7.0 is coming, bringing major performance and hardware upgrades</strong>: Linus Torvalds announced Linux kernel 7.0, following 6.19, with significant updates. Version 6.19 introduces a Live Update Orchestrator for VM-friendly kernel upgrades, encrypted PCIe-to-VM communications, and networking optimizations that can quadruple data transfer speeds. It also improves support for older AMD GPUs by up to 40%. Linux 7.0 will make Rust a permanent language, enable Intel TSX by default for performance gains, and speed up container creation by 40%. These updates, including better Apple Silicon USB-C support and Steam Deck fixes, are set to land in Ubuntu 26.04 LTS, marking a substantial leap in performance and hardware compatibility. <a href="https://app.daily.dev/feed-by-ids?id=4AON6VE6k&amp;id=j4bnStI4p&amp;id=j4bnStI4p&amp;id=yVHM8Igpq&amp;id=tG6Yxobw9&amp;id=tbNmG2QbC&amp;id=umCk0jaGh&amp;id=v9SGa3LyW&amp;id=KRzP2elQI&amp;id=Lh7hAqISa" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI coding agents are changing developer workflows, not eliminating them</strong>: AI coding agents are shifting software development from typing code to defining intent and orchestrating autonomous tasks, with tools like OpenAI’s Codex App and GitHub’s Agent HQ enabling multi-agent collaboration and automated code generation. <a href="https://app.daily.dev/feed-by-ids?id=FEgtdxscE&amp;id=FR00trCGA&amp;id=GOFzDmvEo&amp;id=Ik4RG9gCW&amp;id=l4yo9VPjl&amp;id=l4yo9VPjl&amp;id=2rB0oFTYm&amp;id=X7PELONhI&amp;id=YSL8saeiA&amp;id=aoYvJD5bj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New AI models from Anthropic and OpenAI intensify competition</strong>: Anthropic released Claude Opus 4.6 with a 1 million token context window and multi-agent orchestration, while OpenAI launched GPT-5.3 Codex, a faster agentic coding model, both targeting complex enterprise workflows and code review. <a href="https://app.daily.dev/feed-by-ids?id=qnSEAXH7c&amp;id=qohluZoUs&amp;id=6vLfIe4vv&amp;id=844Q228C9&amp;id=8HMA0hoKL&amp;id=Wbu0BhyXh&amp;id=acBBuIfYQ&amp;id=bheFZPTb3&amp;id=EsMJKZQri&amp;id=cT6pQVJ9R" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft PowerToys proposes a command palette dock</strong>: Microsoft PowerToys is considering a Command Palette Dock, an optional persistent interface element providing one-click access to pinned extensions, customizable and positionable on any screen edge, similar to Linux docks. <a href="https://app.daily.dev/posts/oLV0MsDPD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Amazon Redshift enhances autonomics for multi-cluster environments</strong>: Amazon Redshift extended its automatic optimization features to multi-cluster environments, intelligently managing table layouts and maintenance across all consumer clusters, and now supports allocating extra compute resources for these autonomics to prevent impact on user workloads. <a href="https://app.daily.dev/posts/7o3DKKn6E" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Heroku shifts to sustaining engineering model, prompting migration to alternatives</strong>: Salesforce moved Heroku into a sustaining engineering phase, focusing on stability over new features and ceasing new enterprise contracts, prompting users to consider alternatives like Render, Vercel, DigitalOcean App Platform, or self-hosted solutions like Coolify and Dokku. <a href="https://app.daily.dev/feed-by-ids?id=13PGZg8yI&amp;id=u738eKfMn&amp;id=qAmp5MSPU&amp;id=smfFrov84&amp;id=pW1DiQ44D" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Mistral AI releases Voxtral speech-to-text models with C implementation</strong>: Mistral AI launched Voxtral Transcribe 2, featuring Voxtral Mini Transcribe V2 for batch processing and Voxtral Realtime for live applications, both supporting 13 languages and claiming state-of-the-art accuracy, complemented by an open-source pure C implementation for optimized inference. <a href="https://app.daily.dev/feed-by-ids?id=oP3oVz2Y3&amp;id=PmuBwjJNz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>JetBrains IntelliJ IDEs adopt native Wayland support</strong>: JetBrains IntelliJ-based IDEs will enable native Wayland support by default starting with version 2026.1, addressing long-standing Linux issues like blurry text on HiDPI displays and fractional scaling inconsistencies by using a direct WLToolkit subsystem. <a href="https://app.daily.dev/posts/q5tiuOcND" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong><a href="http://AI.com" target="_blank" rel="noopener nofollow">AI.com</a> domain sold for $70 million: a bet on buzzwords, not brains</strong>: <a href="http://Crypto.com" target="_blank" rel="noopener nofollow">Crypto.com</a> CEO Kris Marszalek bought <a href="http://AI.com" target="_blank" rel="noopener nofollow">AI.com</a> for a cool $70 million, paid in crypto, setting a new domain record. He plans to launch a personal AI agent platform, aiming to compete with OpenAI. Critics are calling it a ‘bet on buzzwords,’ and frankly, spending that much on a domain for a platform that will offer ‘messaging, applications, and stock trading’ sounds less like innovation and more like a desperate attempt to buy relevance. Good luck with that ROI, Kris. <a href="https://app.daily.dev/posts/RoetWzgWa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>X under international scrutiny for grok AI and content moderation: because who needs guardrails?</strong>: French authorities raided X’s Paris office, expanding a criminal investigation into child sexual abuse material, privacy violations, and Holocaust denial, with Elon Musk and Linda Yaccarino summoned for questioning. This follows criticism of Grok AI generating non-consensual sexual deepfakes and child abuse imagery. X’s response? Restricting Grok’s image-editing to paid subscribers and implementing ‘guardrails’ that are, predictably, easily bypassed. It’s almost as if building an AI on a platform known for its content moderation issues was a bad idea. <a href="https://app.daily.dev/posts/i3FTaaYRg" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI bot traffic surges, threatening publishers: because who needs human readers when you have bots?</strong>: AI bots now account for a significant portion of web traffic, with retrieval augmented generation (RAG) bots increasing 33% while human traffic declined 5%. This surge is causing clickthrough rates from AI applications to source websites to plummet to 0.27%, threatening publishers’ financial viability. So, while AI promises to revolutionize information access, it seems it’s mostly just creating an ‘arms race’ where bots fight bots for content that no human will ever click on. Publishers, prepare for your audience to be entirely synthetic. <a href="https://app.daily.dev/posts/5O2tzW5oh" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
