Google announced that Android users can now build apps with AI in Google AI Studio and install them directly on their own device via Google Play, without publishing them to the public store. These private apps won't appear in search or listings and are visible only to their creator, though they must still follow Play's policies. The feature is launching first for users 18 and older in the U.S.
Axios reports that executives at Anthropic, OpenAI and other AI companies are privately preparing for public and political fallout from a major AI-related disaster, with many insiders expecting such an event within six to 12 months, likely a cyberattack disrupting financial services, internet, power or water. Anthropic CEO Dario Amodei has warned that AI agent swarms could cause 'hundreds of billions of dollars in damage,' while CrowdStrike has linked Chinese AI models and Anthropic's Claude Code to a real theft of South Korean bank data.
Google announced a new US feature letting people install personal apps built with AI Studio's vibe-coding tool directly through the Play Store, skipping the need to sideload via a connected computer. These apps remain private and separate from the public store catalog, are tied to a standard Google account, and Google says sharing with select groups is coming soon.
A Show HN project called PDF TO MP3 lets users convert PDF documents into audiobooks by connecting their Codex or Claude account via an MCP server. Document processing is covered by the user's existing AI subscription, while the actual text-to-speech generation is billed separately through an OpenAI or Google Gemini API key. Users authenticate the MCP server, add their speech API key, then attach a PDF in their chat client and request an audiobook conversion.
A developer documented a project called KindleCTL, in which an AI coding agent was used to jailbreak a 10th-generation Kindle Basic and install a custom photo slideshow app. Earlier attempts with other AI models, including GPT-5.6 Sol and Anthropic's Fable 5.1 and Opus 5.5, failed due to technical errors or built-in cybersecurity restrictions, but GPT-6 Astra succeeded in building the jailbreak and the app. The resulting app cycles through ten photos for four minutes each, accessible from the Kindle's library screen alongside books.
A research effort called ELIZAGen is building a family tree of ELIZA chatbot implementations by treating them as biological specimens rather than comparing source code directly. Researchers encode 27 versions spanning 1965-2025 across eight languages using 13 binary architectural features, then analyze the resulting matrix with PAUP*, software standard in evolutionary biology. The work is detailed in a December 2025 draft paper by Jeff Shrager titled 'Reconstructing Program Genealogies from Abstract Program Descriptions: The ELIZA Case Study.'
Meta's Superintelligence Lab released Muse Wonder and Muse Wonder Full, its first frontier-class models, claiming benchmark scores comparable to Fable 5.1 and GPT-6 Astra, including stronger results on coding tasks. Wonder will be priced at $4 per million input tokens and $20 per million output tokens via Meta's Model API, while Wonder Full is reserved for Muse Code subscribers and won't be offered through the API; both will power the Muse agent.
A study covering more than 700 software companies found that AI coding agents raised lines of code by 30 percent, commits by 20 percent, and pull requests by 23 percent, but did not meaningfully increase the rate at which firms resolved tracked features or issues in tools like Jira. Researchers attribute the gap to code review, which took 49 percent longer on average after AI tools were adopted, with more revisions required and more reviewer comments per pull request.
Agent37, a Y Combinator F26 startup, has launched a platform offering sandboxes to run AI agents such as Hermes, OpenClaw and Claude Code at what it says is under 10% of the cost of competing providers. Founder Vishnu says the company's customers average $0.37 per sandbox monthly, and the startup now offers a free tier that costs it roughly $0.05-$0.10 per user per month, with over 90% of new signups on that plan.
Meta has released two new AI models, Muse Wonder and Muse Wonder Full, available immediately to Muse Code subscribers and accessible via the Meta Model API. The company says internal benchmark testing shows the models are competitive with leading systems from OpenAI and Anthropic, though it did not compare them to Google's newest model. Meta plans to eventually use Wonder to power its Muse AI agent, starting with subscribers, though no timeline was given.
A writer reflects on reading Edgar Allan Poe's 1843 short story 'The Gold-Bug' as a middle schooler in India, describing how it introduced them to cryptography through a tale of a buried pirate treasure decoded from symbols. The essay notes Poe wrote the story around 1840, submitted it to a Philadelphia newspaper contest, won $100, and had it published in 1843, making it likely the oldest fiction to feature cryptography.
A developer known as Sword released version 2.01 of PS5SX2, a tool that lets jailbroken PlayStation 5 consoles run PlayStation 2 games from the console's Blu-ray drive. The software dumps disc contents to internal storage before launching titles, and an experimental feature allows loading PS2 game dumps via an external USB Blu-ray drive, potentially extending support to digital-only PS5 models. The tool requires hacked firmware and builds on existing open-source emulation code.