Claude Opus 5 tops ARC & Cyber risk in open models - Hacker News (Jul 25, 2026)

Please support this podcast by checking out our sponsors: - Consensus: AI for Research. Get a free month - https://get.consensus.app/automated_daily - Effortless AI design for presentations, websites, and more with Gamma - https://try.gamma.app/tad - Prezi: Create AI presentations fast - https://try.prezi.com/automated_daily Support The Automated Daily directly: Buy me a coffee: https://buymeacoffee.com/theautomateddaily Today's topics: Claude Opus 5 tops ARC - Anthropic launched Claude Opus 5, a stronger flagship AI model for coding, automation, and knowledge work. Its lead on ARC-AGI-3 also shows that adaptive reasoning remains difficult even for frontier models. Cyber risk in open models - A UK AISI and CAISI assessment says Moonshot AI's Kimi K3 is now the strongest open-weight cyber model they have tested, but it still trails top U.S. systems. The report highlights growing security concerns around accessible AI capability. Android ADB loopback under threat - An Android IssueTracker discussion has raised fears that on-device ADB over loopback could be restricted to Wi-Fi only. That would affect Shizuku, Termux, accessibility tools, and other power-user debugging workflows. Wasmtime turns on Wasm GC - Wasmtime 47 now enables WebAssembly GC and exception handling by default. This is a major step for running higher-level languages on Wasm with smaller binaries and more natural runtime behavior. Playdate gets real-time 3D - A developer built a playable 3D software renderer for the Playdate handheld despite its tiny 1-bit screen and limited hardware. It is a strong example of optimization, smart tradeoffs, and creative game engineering. Hannah Fry wins math prize - Cambridge professor Hannah Fry won the Leelavati Prize for public understanding of mathematics. The award recognizes her work making math accessible through broadcasting, writing, and digital media. Tokyo museum saves dead media - The Extinct Media Museum Tokyo is preserving obsolete devices and recording formats through a hands-on collection. Its open approach helps document media history for researchers, creators, and the public. Apartment aquaponics keeps improving - A small apartment aquaponics project in New York has matured into a stable, compact food-growing system. The update offers practical lessons on urban sustainability, fish care, and low-space gardening. - Android May Restrict On-Device ADB and Break Shizuku Workflows - Anthropic Releases Claude Opus 5 - Hannah Fry Wins International Leelavati Prize for Math Outreach - Developer Builds a 3D Renderer for the Playdate Handheld - ARC Prize Leaderboard Shows Strong Models Still Struggle on ARC-AGI-3 - Kyber Seeks Head of Engineering to Scale AI Document Platform - How an NYC Apartment Aquaponics System Was Improved and Stabilized - Wasmtime 47 Enables WebAssembly GC and Exceptions by Default - UK and U.S. Agencies Assess Kimi K3’s Cyber Capabilities - Extinct Media Museum Tokyo Publishes Official Overview Episode Transcript Claude Opus 5 tops ARC We’ll start with AI. Anthropic has released Claude Opus 5, its new flagship model, and the big pitch is not just raw capability but better efficiency and stronger behavior on long, messy tasks. The company says it is better at coding, debugging, automation, and scientific work, and more reliable when it needs to check its own output instead of giving up early. What makes the timing interesting is that the ARC Prize leaderboard also updated, and Opus 5 currently leads the verified ARC-AGI-3 results. Even so, that top score is only a bit above 30 percent. So the story here is mixed in a useful way: AI systems are clearly improving, but the harder tests still show how far they are from flexible, human-like adaptation. Cyber risk in open models Staying with AI, there is also a new warning sign in cybersecurity. A preliminary assessment from UK AISI and CAISI looked at Moonshot AI’s Kimi K3 and found that it is now the most capable open-weight cyber model they have tested so far. It still falls short of the leading U.S. frontier models, but the gap is narrowing enough to matter. In practical testing, it made real progress on exploit tasks and simulated network attacks, even if it was not yet top tier. Why this matters is simple: when stronger cyber capability becomes available in more open models, the conversation shifts from hypothetical risk to operational risk. Android ADB loopback under threat On the Android side, a blog post is drawing attention to a possible future restriction on ADB connections made directly on-device. This is not an official Google announcement, but rather a concern based on an ongoing IssueTracker discussion tied to a security flaw. The worry is that ADB could end up limited to the main Wi-Fi interface, which would break loopback-based workflows used by tools like Shizuku, Termux setups, and other local debugging methods. The author’s point is that these are not fringe abuse cases. For many developers and power users, they a