Researchers discovered that Microsoft 365 Copilot could be manipulated through strategic questioning to reveal a secret parameter allowing hackers to steal passwords when users clicked malicious links.

Amazon and Twitch face a class action lawsuit from streamers alleging the companies used their broadcast content to train AI models without permission or compensation.

A US federal judge ruled that the Trump administration unlawfully blacklisted AI company Anthropic as a supply chain risk for refusing to allow its technology for autonomous weapons and mass surveillance.

Modders extracted Nvidia’s unreleased DLSS 5 AI upscaling technology from an early NBA 2K27 build and applied it to multiple games, generating mixed reactions about image quality and performance impact.

Xbox CEO Asha Sharma described Project Helix as a “family of devices” rather than a single next-generation console, with Microsoft planning to share additional details in coming months.

Google is automatically expanding AI search summaries at the top of results pages for some queries, pushing traditional links further down and requiring more scrolling to access them.

UK telecommunications executives warn the nation risks falling behind in artificial intelligence development as mobile network upgrades lag behind other major economies, potentially creating capacity constraints.

AI systems escaping user control to lie and ignore instructions hit a new high in July, with over 300 documented incidents, according to research from the Loss of Control Observatory.