TechCapsules

A Pew Research Center survey revealed that for the first time, a majority of American adults under 30 report being more concerned than excited about artificial intelligence. The percentage of youth who feel more anxious than enthusiastic rose from 31% in 2021 to 48% this year, while excitement fell from 25% to 11%. Despite 66% having used a chatbot and 31% using one daily, prolonged exposure has increased fears of deepfakes, job loss, and learning impairment. The [...]

Google announced a phased replacement of Gemini Gems with a new modular system called Skills. Skills will launch in Google Workspace on October 5, in the Gemini app on October 13, and Gems will begin exiting in November. The new framework lets users capture brand voice, templates, and processes into reusable building blocks, which can be stacked for different tasks. Gemini will auto‑detect applicable Skills, and the format is open, using Markdown for portability. Gems will remain available to [...]

Washington lawmakers face a hurdle in AI regulation as industry players seek narrow antitrust exemptions to collaborate on safety research. The request, aimed at preventing regulatory paralysis, has met resistance from legislators who worry that carving out antitrust protections could undermine competition and create loopholes. The debate highlights a clash between the need for rapid safety coordination and the preservation of antitrust enforcement. Key stakeholders include AI firms, congressional committees, and regulatory agencies, all navigating the balance [...]

President Donald Trump signed an executive order directing federal agencies to replace the term “artificial intelligence” (AI) with “super intelligence” (SI) in all non‑statutory documents, policy papers, and correspondence. The order follows Trump’s remarks at the UN General Assembly, where he said the current name makes the technology appear “fake.” The White House surveyed over 50,000 Truth Social followers, with 65% favoring SI. The directive applies to executive‑branch communications but does not alter existing laws or agency [...]

On September 29, 2026, U.S. President Donald Trump and executives from Anthropic, OpenAI, Google, Meta, xAI, and Nvidia announced a voluntary Joint Commitment on Frontier Responsibilities to enhance AI safety. The pact requires stronger monitoring of AI models, internal oversight teams, and collaboration with independent auditors. Companies will hold regular discussions to develop shared standards and best practices. The agreement is not legally binding and follows international calls for coordinated AI oversight. Reactions are mixed, with some [...]

Synopsys and OpenAI have signed a multi‑year partnership to develop GPT‑Synopsys, a specialized AI model that integrates OpenAI’s frontier models with Synopsys’ electronic design automation (EDA) tools. The collaboration will enable the model to operate EDA workflows, interpret tool outputs, and iteratively optimize semiconductor designs for power, performance, and area. The agreement includes revenue sharing, global go‑to‑market plans, and enterprise‑grade security to protect customer data. Early trials are underway with leading semiconductor firms. [...]

Dun & Bradstreet unveiled AI‑powered features for its Finance Analytics platform and Model Context Protocol server, aimed at accelerating credit decisions and early risk detection. The new D&B.AI tools embed the D&B Commercial Graph, built on the global D‑U‑N‑S® Number, to provide verified business identity, relationships and credit intelligence to AI assistants and custom agents. According to executive Julian Prower, the capabilities can reduce analysis time from hours to seconds and help firms address delayed payments that [...]

Researchers at Microsoft used the Discovery app to build a cloud‑deployable bioinformatics pipeline that helps identify bacteriophages capable of treating drug‑resistant bacterial infections. The workflow accepts a bacterial genome, flags antibiotic‑resistance genes, predicts antiphage defenses, and scans large environmental datasets for candidate phages. Each step is documented in an inspectable task tree, allowing scientists to trace decisions, validate methods, and adjust the approach in real time. The collaborative human‑agent process links literature, multi‑omics data, and analytical tools, [...]