<h2>Must know</h2>
<ul>
<li><strong>Ladybird browser adopts rust with AI help</strong>: The Ladybird browser project is ditching C++ for Rust, primarily for memory safety. After a year lost trying Swift, 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. The initial Rust port of LibJS, their JavaScript engine, achieved byte-for-byte identical output with zero regressions, though the C++ development continues as the main track. This highlights how AI can accelerate large-scale refactoring efforts, even if it’s a human-directed process. <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>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. What’s interesting is that this technically unsophisticated group used generative AI tools for every phase of the attack, from reconnaissance to generating exploitation instructions and writing scripts. This shows how AI is lowering the bar for cyberattacks, enabling less skilled actors to operate at a larger scale, though AWS notes basic security hygiene would have prevented most of these incidents. <a href="https://app.daily.dev/posts/azBzuZltn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ZeroDayRAT mobile spyware emerges, bypassing MFA</strong>: A new mobile spyware, ZeroDayRAT, is being openly sold on Telegram, targeting Android 5-16 and iOS up to 26 for $2,000. It offers capabilities like keylogging, SMS control (bypassing MFA), screen recording, and credential theft, essentially commoditizing nation-state-level surveillance. This poses significant risks, especially for enterprises with BYOD policies, as it requires no technical expertise for attackers and can be distributed via smishing or fake app stores. <a href="https://app.daily.dev/posts/Q0yMWpjwU" 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, accessible with a single subscription and allowing users to run and compare multiple agents simultaneously. <a href="https://app.daily.dev/posts/CxN7d4Xat" 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. <a href="https://app.daily.dev/posts/NOSp6Xv2e" 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, designed to run on mobile devices and serve languages often overlooked by larger models. <a href="https://app.daily.dev/posts/QusoaclzE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI responses API websockets integration and performance impact</strong>: OpenAI’s Responses API now supports WebSockets, enabling AI agents to run 30-40% faster, particularly for frequent tool calls, with Cursor editor already upgrading users for up to 30% speed improvements. <a href="https://app.daily.dev/posts/CNQCNZ2N0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Firefox 148 release with AI controls</strong>: Firefox 148 introduces an ‘AI kill switch’ allowing users to disable all AI-powered features or selectively block individual ones, with Mozilla promising future updates won’t re-enable them. <a href="https://app.daily.dev/posts/UNAL37VrS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub enterprise cloud extends IP allow list to user namespaces</strong>: GitHub Enterprise Cloud now extends IP allow list configuration to user namespaces for Enterprise Managed Users (EMUs), restricting access to web UI, git protocol, and API access based on the enterprise’s IP allow list. <a href="https://app.daily.dev/posts/5t2T6mrR4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Redesigning a homepage in 20 minutes: Figma MCP + Atlassian Design System MCP + Rovo Dev CLI</strong>: A new workflow demonstrates redesigning a React homepage in 20 minutes by chaining Figma MCP, Atlassian Design System MCP, and Rovo Dev CLI to extract design, retrieve component APIs, and generate production-ready, ADS-compliant code, including fixing accessibility violations. <a href="https://app.daily.dev/posts/ZWlO975fr" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic accuses chinese AI labs of industrial-scale model distillation</strong>: Anthropic is publicly accusing Chinese AI labs of “industrial-scale distillation attacks” on its Claude models, claiming they created 24,000 fraudulent accounts to extract capabilities. This, coming from a company facing its own copyright lawsuits for training on scraped content, feels a bit like the pot calling the kettle black, conveniently aligning with calls for stricter chip export controls. Funny how “national security threat” gets thrown around when it suits the narrative, isn’t it? <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>Anthropic’s claude code security launch and market reaction</strong>: Anthropic launched Claude Code Security, an AI tool for vulnerability detection, which apparently caused cybersecurity stocks to tank. Analysts called the market reaction “illogical,” and frankly, they’re probably right. While Anthropic claims it found 500+ vulnerabilities, they haven’t bothered to publish false-positive rates or cost data. So, is it a breakthrough or just another AI marketing blitz designed to make us think our jobs are obsolete? <a href="https://app.daily.dev/posts/bgcFYmVK1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google deepmind claims image generation breakthrough</strong>: Google DeepMind supposedly “just solved” a “dirty problem” in image generation with an “embarrassingly elegant” fix, according to some random retweet. With zero actual details or official announcements, this sounds less like a breakthrough and more like someone trying to generate hype for their social media feed. I’ll believe it when I see it, and by “it,” I mean actual technical specifics, not just breathless adjectives. <a href="https://app.daily.dev/posts/fzJvtM3B4" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
