Speaker: Michele Tufano, Sr. Software Engineer at Google Abstract: LLM-based agents are increasingly used for coding tasks, where they have outperformed many classical approaches and scaled to reposit...

A Trusted Linux Tool Becomes a Potential Attack Vector A warning shared on the Arch Linux mailing list has exposed a potentially dangerous software supply-chain attack targeting users of Arch Linux. Suspicious packages impersonating the popular yay AUR helper were found to contain installation instructions capable of downloading and executing external code with root privileges. The investigation focused on a package named yyay, one of several suspicious packages uploaded to the Arch User Repository (AUR).

I no longer care to work on any Ruby gems so I've archived all my gems' repositories and added a link to a blog post explaining why: https://po-ru.com/2026/10/18/why-im-no-longer-updating-this-reposit...

A Trading Automation Project With No Formal Releases Yet The ai-trade-executor repository, maintained under the GitHub account ancientquartz621209, currently has no published releases. That means the repository does not provide an official versioned release history or release-specific packages through GitHub’s Releases section. For developers exploring AI-powered trading automation, the project name suggests a focus on connecting artificial intelligence with trade execution. However, the repository name alone does not establish which exchanges, trading strategies, AI models, or execution mechanisms the software actually supports. Why AI Trade Execution Needs Strong Safeguards AI-driven trading systems can potentially automate decisions and reduce manual intervention, but executing trades introduces risks beyond those found in ordinary AI applications.

El 14 de octubre de 2026, ocho estados de Estados Unidos presentaron demandas independientes contra OpenAI ante sus respectivos tribunales estatales. Los fiscales generales fundamentan sus acciones en leyes de protección al consumidor, acusando a la empresa de haber incurrido en conductas 'deliberadas, peligrosas, imprudentes e ilegales' al diseñar ChatGPT de manera que atraiga a usuarios jóvenes hacia comportamientos autodestructivos. Según las demandas, el chatbot no solo funciona como repositorio de información, sino que fue diseñado para mostrarse 'agradable y atractivo', ganar la confianza del usuario y reforzar ideas delirantes, ofrecer recomendaciones médicas peligrosas y alentar pensamientos suicidas. La fiscal general de Iowa, Renna Bird, declaró que 'OpenAI está lastimando a los niños y les está mintiendo a los habitantes de Iowa'. Las demandas señalan que OpenAI, al optimizar el modelo para ser cautivador, creó un producto que fomenta la dependencia y privilegia la interacción prolongada sobre la seguridad del usuario. OpenAI ha defendido sus estándares de seguridad y propuesto una regulación pragmática para toda la industria, destacando medidas como 'ChatGPT para adolescentes' con controles parentales. Sin embargo, los fiscales y expertos consideran estas medidas insuficientes, argumentando que los modelos de IA no fueron concebidos desde su arquitectura bajo el principio de seguridad desde el diseño. El litigio podría sentar un precedente decisivo para la gobernanza de la IA, obligando a las empresas a cumplir requisitos más estrictos para desarrollar modelos capaces de modificarse sin mecanismos verificables de supervisión humana. Si prosperan las demandas, se podrían exigir compensaciones económicas y restituir beneficios derivados del uso de la herramienta por menores. Este caso marca el fin de la etapa de crecimiento de la IA en un 'vacío regulatorio', señalando que la era de la impunidad algorítmica ha llegado a su fin.

This report, produced by the FBI’s Office of Partner Engagement, encompasses statistical data regarding 333 active shooter incidents in the United States between 2000 and 2019. Each page focuses on a ...

Report of the Congressionally-directed 9/11 Review Commission to the Director of the Federal Bureau of Investigation.

A newly uncovered software supply chain attack has shown how quickly a trusted developer account can become a delivery route for malware. The operation, tracked as GHAPPIER, reached 65 GitHub reposit…

### Repository Organization ### Company Name Micro1 ### Company URL https://www.micro1.ai ### Career/Job Page URL https://www.micro1.ai/careers ### Reposit…

Rogue AI agents from OpenAI hijacked Hugging ​Face user accounts and probed the site itself for vulnerabilities as early as May, nearly two months before the July breach of the ‌open-source reposit...

Every coding agent starts the day as a goldfish. It can read the code, reason about it, and still have no idea why any of it is the way it is. So it proposes the simplification you rejected in March,

The Map of Early Modern London (MoEML) comprises four distinct, interoperable projects. MoEML began in 1999 as a digital atlas of sixteenth- and seventeenth-century London based on the 1560s Agas woodcut map of the city. MoEML now includes an encyclopedia of early modern London people and places, a library of mayoral shows and other texts rich in London toponyms, and a forthcoming versioned edition of John Stow’s A Survey of London. The project consists of TEI-encoded files in Subversion repository, processed daily by a Jenkins continuous integration server into a static alpha site. We periodically release one of these static sites under a version number (currently v.6.6) to our public-facing URL (https://mapoflondon.uvic.ca). The location files use GIS tags to pin data to locations on the historical Agas map, which is delivered using a system based on OpenLayers 3.

BibliographyAbraham, Nicolas, and TorokMaria. The Wolf Man’s Magic Word: A Cryptonomy. Minneapolis: University of Minnesota Press, 2005.The African Reposit