<h2>Must know</h2>
<ul>
<li><strong>Ladybird browser pivots to rust with AI assistance</strong>: The Ladybird browser project is ditching Swift for Rust, primarily for memory safety, after a year of lost progress. Lead developer Andreas Kling used AI agents like Claude Code and Codex to translate 25,000 lines of C++ to Rust in just two weeks, a task estimated to take months manually. This human-directed AI workflow resulted in byte-for-byte identical output and zero test regressions, showing AI’s potential for accelerating large-scale refactoring, as reported by multiple sources. <a href="https://app.daily.dev/feed-by-ids?id=bavH6r0ZC&amp;id=1QrlHS5ue" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic accuses chinese labs of industrial-scale model distillation</strong>: Anthropic claims Chinese AI labs DeepSeek, Moonshot AI, and MiniMax created over 24,000 fraudulent accounts to distill its Claude models, generating 16 million exchanges to extract capabilities like agentic reasoning and tool use. Anthropic frames this as a national security threat, arguing distilled models lack safety guardrails and could be misused, conveniently aligning with ongoing US debates on AI chip export controls. This comes shortly after OpenAI made similar claims. <a href="https://app.daily.dev/feed-by-ids?id=YcZmCF704&amp;id=s76VxSBCH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-augmented fortigate firewall compromises reported by AWS</strong>: AWS detected a financially motivated, Russian-speaking threat actor compromising over 600 FortiGate firewall instances across 55+ countries. This unsophisticated group used generative AI tools for reconnaissance, exploitation instructions, scripting, and operational notes, demonstrating how AI can enable less skilled attackers to operate at scale. The attacks leveraged exposed management interfaces, weak credentials, and lack of MFA, highlighting the continued importance of basic security hygiene. <a href="https://app.daily.dev/posts/azBzuZltn" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>GitHub introduces agent HQ for unified AI coding agents</strong>: GitHub announced Agent HQ, a unified platform for coding agents like Claude, Codex, and GitHub Copilot, allowing users to run and compare multiple frontier agents simultaneously with a single Copilot Pro Plus or Enterprise subscription. <a href="https://app.daily.dev/posts/CxN7d4Xat" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes 1.35 strengthens AI/ML workload support</strong>: Kubernetes v1.35 (‘Timbernetes’) introduces features like alpha gang scheduling for distributed training, stable in-place Pod resource resize for inference, and continued Dynamic Resource Allocation progress, solidifying its role as an AI/ML operating system. <a href="https://app.daily.dev/feed-by-ids?id=mAyDHBOpw&amp;id=nytQTToPj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI releases gpt-realtime-1.5 model</strong>: OpenAI released gpt-realtime-1.5 via its Realtime API, designed to strengthen voice workflows with improved instruction following, enhanced tool calling, and better multilingual accuracy for low-latency, multimodal interactions. <a href="https://app.daily.dev/posts/6I1BavGOr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Uber open-sources uForwarder kafka consumer proxy</strong>: Uber open-sourced uForwarder, a push-based Kafka consumer proxy designed to handle trillions of messages daily, centralizing offset management and simplifying consumer logic for microservices with features like context-aware routing and adaptive auto-rebalancing. <a href="https://app.daily.dev/posts/NOSp6Xv2e" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ZeroDayRAT mobile spyware emerges, bypassing MFA</strong>: ZeroDayRAT, a new mobile spyware sold on Telegram for $2,000, targets Android 5-16 and iOS up to 26, offering capabilities like keylogging, SMS control (for MFA bypass), screen recording, and credential theft, commoditizing nation-state-level surveillance. <a href="https://app.daily.dev/posts/Q0yMWpjwU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cohere releases aya tiny multilingual models</strong>: Cohere released the Aya Tiny model family, small (~3.3B parameter) multilingual language models supporting over 70 languages, including low-resource ones, designed for mobile devices and specialized for different global regions. <a href="https://app.daily.dev/posts/QusoaclzE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenClaw beta release with security patches and new features</strong>: OpenClaw released a new beta version focusing on security patches, bug fixes, and regression fixes, while also adding Kilo provider, Kimi vision, and video support. <a href="https://app.daily.dev/posts/EsafTF3ga" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Google DeepMind claims image generation breakthrough</strong>: Some Twitter user is hyping up Google DeepMind for ‘just solving’ a ‘dirty problem’ in image generation with an ‘embarrassingly elegant’ fix. No details, no official announcement, just pure, unadulterated social media buzz. Sounds like another day in AI land where vague claims are the new concrete achievements. <a href="https://app.daily.dev/posts/fzJvtM3B4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI updates AI coding evaluation standard</strong>: OpenAI is now recommending SWE-bench Pro for AI coding evaluations because their previous benchmark, SWE-bench Verified, became ‘saturated’ due to ‘test-design issues and contamination.’ So, they’re moving the goalposts after their models got too good at the old game. Nothing to see here, just continuous improvement, definitely not a convenient way to keep the ‘progress’ narrative going. <a href="https://app.daily.dev/posts/oZGyDrjcX" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s claude code security launch triggers ‘illogical’ market reaction</strong>: Anthropic launched Claude Code Security, an AI tool for vulnerability detection, and apparently, the stock market had a meltdown, with cybersecurity stocks plummeting. Analysts are calling the reaction ‘illogical,’ which is a polite way of saying investors are panicking over an AI tool that’s still in ‘limited research preview’ and has yet to publish false-positive rates or cost data. Classic market overreaction to shiny new tech. <a href="https://app.daily.dev/posts/bgcFYmVK1" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
