Every story in this issue with a short summary, so you can decide what to read before leaving the site. Summaries are written by AI (Claude) only from each publisher's own description and checked by Jev for anything not in that description. They are not full translations. Follow "Read the original" for the full story.
The article carries the quote: "With today's science we usually can't show with high confidence that dangerous behavior isn't there."
Read the original →Jason Tucker, Virginia Dignum and Petter Ericson raise concerns about the first 'Thematic Brief' from the UN Independent International Scientific Panel on AI.
Read the original →New UNSW research finds that AI chatbots can be manipulated to bypass safety guardrails. The findings raise AI safety questions for schools.
Read the original →Researchers developed an AI-assisted inverse-design approach that identifies multiple synthesis routes to carbon nanotube arrays with closely matched height, density and alignment. Experiments across three targets showed that similar array structures can emerge from different processing conditions, while individual nanotubes keep microscopic differences. The findings offer more flexibility for controlled nanomaterial synthesis.
Read the original →Hackers used the Chinese open-source AI agent Artex to breach seven South Korean banks. They stole data on 68,000 people.
Read the original →Jason Kwon, one of OpenAI's most senior executives, came from San Francisco to appear before a federal parliamentary inquiry. The inquiry followed AI agents infiltrating sensitive government information.
Read the original →AWS invests in tools that help customers align with international standards for responsible AI governance. The post explains the AI system impact assessment, how it improves enterprise-wide risk management, and how ISO/IEC 42005:2025 codifies best practices for conducting and documenting these assessments.
Read the original →The complaint was filed Friday in Illinois federal court.
Read the original →Courts are increasingly being asked how AI tools may be used with confidential discovery materials as generative AI becomes more common in litigation workflows. A recent District of Colorado decision distinguishes between public consumer AI tools and enterprise platforms with specified safeguards. It also addresses proposed provisions requiring approval before new AI tools may be used.
Read the original →