<h2>Must know</h2>
<ul>
<li><strong>Ladybird browser pivots to rust with AI assistance</strong>: The Ladybird browser project is ditching C++ for Rust, primarily for memory safety. After a year lost trying Swift, lead developer Andreas Kling announced the pivot, using AI agents like Claude Code and Codex to translate 25,000 lines of C++ to Rust in just two weeks. This human-directed AI workflow, involving hundreds of prompts and adversarial reviews, achieved byte-for-byte identical output with zero regressions, proving AI’s potential for accelerating large-scale refactoring, as reported by Changelog News. <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>Linus torvalds drops old linux kconfig option to address kernel log spam</strong>: Linus Torvalds removed the WARN_ALL_UNSEEDED_RANDOM Kconfig option from the Linux kernel, citing excessive log spam that obscured real bug reports. This option, meant to flag unseeded random number generator calls, was generating so much noise it made debugging difficult. Torvalds recommends using boot-time tracing for those who genuinely need to investigate unseeded RNG calls, emphasizing the importance of clear, actionable logs over verbose, unhelpful warnings, according to Phoronix. <a href="https://app.daily.dev/posts/uxqwbGYRL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic accuses chinese ai labs of industrial-scale model distillation</strong>: Anthropic is publicly accusing Chinese AI labs DeepSeek, Moonshot AI, and MiniMax of “industrial-scale distillation attacks” on its Claude models. They claim these labs created over 24,000 fraudulent accounts and generated 16 million exchanges to extract Claude’s capabilities, framing it as a national security threat. This comes amidst ongoing US debates on AI chip export controls, with Anthropic using these accusations to push for stricter restrictions, despite facing copyright infringement lawsuits themselves for training on scraped content. <a href="https://app.daily.dev/posts/YcZmCF704" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Anthropic launches claude code security</strong>: Anthropic launched Claude Code Security, an AI-powered tool for vulnerability detection and patch suggestion, which caused a stir in cybersecurity stock markets, though analysts called the selloff “illogical.” <a href="https://app.daily.dev/posts/bgcFYmVK1" 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>GitHub introduces agent hq for unified ai coding agents</strong>: GitHub announced Agent HQ, a unified platform for coding agents that integrates Claude, Codex, and GitHub Copilot, allowing users to run and compare multiple frontier agents simultaneously with a single subscription. <a href="https://app.daily.dev/posts/CxN7d4Xat" 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 through its Realtime API, aiming to strengthen voice workflows with improved instruction following, enhanced tool calling, and better multilingual accuracy for low-latency speech-to-speech interactions. <a href="https://app.daily.dev/posts/6I1BavGOr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-augmented fortigate firewall compromises reported by aws</strong>: AWS reported that an unsophisticated, financially motivated threat actor compromised over 600 FortiGate firewall instances across 55+ countries using generative AI tools for reconnaissance and exploitation, leveraging basic security hygiene failures. <a href="https://app.daily.dev/posts/azBzuZltn" 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 designed to support over 70 languages, including low-resource ones, and capable of running on mobile devices. <a href="https://app.daily.dev/posts/QusoaclzE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ZeroDayRAT mobile spyware emerges, bypassing mfa</strong>: ZeroDayRAT, a new mobile spyware family sold openly on Telegram, targets Android and iOS, offering capabilities like keylogging, SMS control, and credential theft, enabling MFA bypass and commoditizing nation-state-level surveillance. <a href="https://app.daily.dev/posts/Q0yMWpjwU" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>OpenAI updates ai coding evaluation standard</strong>: OpenAI, in a move that absolutely no one saw coming, has decided its previous benchmark, SWE-bench Verified, is “saturated” and “contaminated.” So, naturally, they’ve introduced SWE-bench Pro, which they assure us is much more rigorous. Because when your own benchmark makes your models look too good, the obvious solution is to invent a new one, as reported by OpenAI/OpenAIDevs. <a href="https://app.daily.dev/posts/oZGyDrjcX" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI responses api websockets integration and performance impact</strong>: Steven Heidel (RT) announced that OpenAI’s Responses API now supports WebSockets, claiming AI agents will run 30-40% faster. Cursor, an editor, also jumped on the bandwagon, claiming their OpenAI models are now up to 30% faster. It’s almost as if adding a standard, decades-old web technology suddenly unlocks magical AI speed, according to Steven Heidel (RT) and leerob (RT). <a href="https://app.daily.dev/posts/CNQCNZ2N0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google deepmind claims image generation breakthrough</strong>: Google DeepMind has reportedly “just solved one of the dirtiest problems in image generation” with an “embarrassingly elegant” fix, according to a highly enthusiastic tweet from @rryssf_. Details? Specifics? Actual evidence? Who needs 'em when you have such compelling adjectives and a retweet to spread the hype? <a href="https://app.daily.dev/posts/fzJvtM3B4" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
