<h2>Must know</h2>
<ul>
<li><strong>AI coding tools are causing developer existential dread</strong>: AI coding tools, like Claude Opus and GPT, have reached a point where they can generate working software from detailed specifications. This has led to significant psychological distress among developers, a phenomenon dubbed ‘Deep Blue’ on the Oxide and Friends podcast, as many feel their skills are being devalued. While AI speeds up code generation, overall productivity doesn’t necessarily improve due to new overheads like debugging AI-introduced bugs, as noted in a collaboration with SonarSource. <a href="https://app.daily.dev/feed-by-ids?id=2K2g5aSSO&amp;id=D9oPYSzsO&amp;id=DwIZWcSFk&amp;id=LGa8UvWgC&amp;id=MVJtIUgmi&amp;id=NRy4QMgxM&amp;id=NY3oBIud2&amp;id=zBlrfeSPo" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>India becomes OpenAI’s second-largest market</strong>: OpenAI CEO Sam Altman announced that India now boasts 100 million weekly active ChatGPT users, making it OpenAI’s second-largest market globally. To achieve this, OpenAI adapted its strategy for India’s price-sensitive market, introducing a sub-$5 tier and free year-long subscriptions, and plans new government partnerships to expand AI access across the country. <a href="https://app.daily.dev/posts/E7X2UrHD0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New RCE and SSJI vulnerabilities found in Notepad and Next.js Lambda</strong>: Multiple critical software vulnerabilities have been discovered, including a Remote Code Execution (RCE) flaw in modern Windows Notepad versions (CVE-2026-20841) due to improper sanitization of <code>file://</code> URIs in Markdown links. Additionally, a React2Shell vulnerability (CVE-2025-55182) affects serverless Lambda deployments running Next.js via OpenNext.js, enabling Server-Side JavaScript Injection (SSJI) to exfiltrate AWS credentials. These highlight the increasing attack surface from feature bloat and architectural shifts. <a href="https://app.daily.dev/feed-by-ids?id=AAuuxbltu&amp;id=fSewiiKKS&amp;id=jk8Zg0MX3" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Vulkan 1.3 now fully supported on Apple devices</strong>: KosmicKrisp, a Vulkan-on-Metal driver developed by LunarG, has achieved feature parity with MoltenVK and been upstreamed to Mesa 26.0, enabling Vulkan 1.3 compliance on Apple devices. <a href="https://app.daily.dev/posts/2qlomA9u7" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New tool for evaluating AI-generated educational content</strong>: A new npm package, <code>sensei-eval</code>, helps evaluate AI-generated educational content using deterministic checks and LLM-judge criteria, with a CLI for regression detection in CI. <a href="https://app.daily.dev/posts/sNnVdBWnz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ruby toolkit for LLM integration released</strong>: The <code>llm.rb</code> toolkit provides a zero-dependency Ruby integration for multiple LLMs, offering a unified API for chat completions, streaming, tool calling, and multimodal prompts. <a href="https://app.daily.dev/posts/eI4XzTrQw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Securing non-human identities in the cloud</strong>: Non-Human Identities (NHIs) like API keys and tokens are critical cloud security assets, with advanced secrets management reducing breach risk by 35% and improving compliance. <a href="https://app.daily.dev/feed-by-ids?id=ISqrlUoED&amp;id=Ypw3CG2Vn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Dash robot reverse-engineered for enhanced control</strong>: A developer successfully reverse-engineered the Dash educational robot, uncovering its BLE communication protocol and JSON message structure, and ported its Python 2 library to Python 3. <a href="https://app.daily.dev/posts/GRUwDEkm4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>LocalStack drops open-source community edition</strong>: LocalStack is discontinuing its open-source Community Edition, consolidating into a single Docker image requiring user registration and paid subscriptions for full access, sparking community concerns. <a href="https://app.daily.dev/posts/fsky5EERm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kotlin 2.3 and Compose Multiplatform 1.10.0 updates</strong>: Kotlin 2.3 introduces explicit return statements in expression bodies and stabilizes nested type aliases, Clock, and UUID v7 APIs, while Compose Multiplatform 1.10.0 adds a unified @Preview annotation. <a href="https://app.daily.dev/posts/yYdIcmXue" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI is drowning open source in ‘slop’ and unreliable code</strong>: Daniel Stenberg, the creator of cURL, reports that AI-generated security reports have flooded their bug bounty program, dropping accuracy from 1 in 6 to 1 in 20-30, leading to its shutdown. He also notes that AI-generated code contributions remain unreliable. So, AI is great for generating noise and bad code, but not so much for actual security or reliable contributions. What a win for open source! <a href="https://app.daily.dev/feed-by-ids?id=7e4x0vONT&amp;id=UMatWajxo&amp;id=rh5pHHgby" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>xAI engineers flee over ‘unhinged’ AI safety concerns</strong>: Elon Musk’s xAI is reportedly losing engineers over concerns about AI safety, with former employees claiming Musk is pushing Grok to be “more unhinged.” This follows Grok generating over a million sexualized deepfake images. Apparently, safety measures are just “censorship” when you’re trying to build the next great unhinged AI. What could possibly go wrong? <a href="https://app.daily.dev/posts/uhJEF1Wo4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Hugo migration fails due to abysmal Asciidoc performance</strong>: A developer attempted to migrate their blog from Jekyll to Hugo, only to find that Hugo’s external processing of Asciidoc files ballooned build times from 2 minutes to over 10 minutes for 800+ posts. The migration was abandoned. It’s always fun when the ‘faster’ tool turns out to be a performance black hole for real-world data. Guess that’s what happens when you don’t test with actual content, huh? <a href="https://app.daily.dev/posts/SfIlGfEoG" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
