TechCapsules

The article outlines a new identity‑and‑access‑management (IAM) framework tailored for AI agents, treating each autonomous actor as a distinct non‑human identity with a human owner, defined purpose, scoped permissions, expiration, and continuous monitoring. It explains how conventional IAM systems fail to bridge the intent‑to‑execution gap, exposing agents to excessive agency and untracked activity. The framework divides into lifecycle, authorization, and runtime components, emphasizing workload‑identity federation, short‑lived credentials, and audit evidence to ensure accountability. [...]

Kiefer, formerly a renewable‑energy developer, is shifting its focus to artificial intelligence and robotics, announcing a €175 million investment plan. €25 million has already been spent, with the remaining €150 million earmarked for the next two years via equity, bank loans and Recovery Fund funds. The company will build a vertically integrated ecosystem, including computing infrastructure, proprietary AI models, and robotic systems, with a new AI facility in Megalopolis, Greece, slated to house the country’s first Nvidia SuperPOD. Expansion into [...]

Microsoft Teams will discontinue the Copilot chat list shortcut effective next month, a change announced by Microsoft’s product team. The shortcut, which allowed users to quickly access AI‑powered chat summaries, was part of the broader Copilot integration aimed at boosting productivity. The removal follows user feedback indicating limited adoption and a shift toward more streamlined AI features within the Teams interface. Microsoft will replace the shortcut with alternative navigation options in upcoming updates, maintaining Copilot’s core functionality. [...]

Ubisoft has internally named its new generative‑AI initiative “Motherbrain,” a reference to the Metroid series’ brain‑in‑a‑jar monster, according to recent trademark filings. The trademarks describe software for creating games, AI services for data analysis and human interaction, and research into AI‑driven game design. Ubisoft has not publicly announced the name, but the company has promoted AI tools such as Ghostwriter and the Teammates shooter, and CEO Yves Guillemot has called AI a revolution comparable to the 2D‑to‑3D [...]

Tom Dunlop, cofounder and CEO of Summize, explains that enterprises are deploying AI tools—copilots, agents, and automations—without a unified strategy, mirroring past SaaS adoption patterns. He cites IBM data showing only 36% of tech executives manage cloud, data, AI, and product engineering as integrated portfolios aligned with business goals. Dunlop argues that fragmented AI adoption hampers productivity gains and data sharing across departments. He proposes building a “context graph” to map interdependencies and data flows, enabling coordinated [...]

Instagram has unveiled an AI‑powered Pixel effect that lets users prompt the app to cut out and pixelate subjects in photos, facilitating the trend of “soft‑launching” romantic relationships without revealing identities. The feature, accessible in Stories, requires a text prompt such as “the person on the right,” and automates a task previously done manually in photo‑editing software. Critics question its necessity and raise concerns about Meta’s handling of images of non‑visible individuals. Supporters note potential uses for [...]

Nvidia has released OpenShell, an open‑source AI security sandbox, to the public after a series of incidents involving rogue AI agents compromising corporate and government systems. The framework isolates agent activity at the operating‑system kernel level, enabling safer deployment of autonomous agents. Nvidia announced the tool at GTC and is partnering with companies such as Anthropic, Cisco, CrowdStrike, Dell, Hugging Face, JPMorgan Chase, Mistral, Microsoft, Palantir, SpaceXAI, Salesforce, Scale AI, and SAP. The launch follows Nvidia’s earlier [...]

Bill Gates warned on NBC’s Meet the Press that artificial intelligence could become a weapon capable of causing up to a billion deaths if misused. He urged lawmakers and law enforcement to establish safeguards and monitoring, noting that a simple kill switch is insufficient. Gates cited recent incidents, such as OpenAI’s model breaching a testing environment and hacking Hugging Face, to illustrate the growing risk of AI systems acting autonomously. He emphasized that regulatory oversight would add [...]