Desktop agents expand as privacy concerns grow
Anthropic reportedly debuted on Nasdaq in a record-setting IPO, with CEO Dario Amodei reiterating commitments around the pace of AI development. Its substantial computing commitments also intensified scrutiny of circular financing across the industry. Separately, reports linked Jensen Huang’s resignation as Nvidia CEO to an alleged relationship with a woman suspected of spying for China. Interim CEO Colette Kress sought to reassure investors amid share-price volatility.
Google added direct AI conversations to its search homepage and computer-use capabilities to Gemini on macOS, enabling interaction with local files, desktop applications and multistep workflows. Apple is reportedly building an agent capable of browsing, operating across applications and writing code. Meta’s Muse faces reported privacy problems involving mixed-up conversations and exposure of personal requests, potentially complicating plans for its accompanying wearable hardware.
Other updates focused on usability and developer infrastructure. OpenAI introduced live dictation for ChatGPT and detailed Codex-powered workflows for identifying and fixing performance problems. Google added tool retrieval to the Gemini API, while Superhuman extended its Go assistant into documents and email. Canva introduced an option to disable all generative AI features. Research developments included Anthropic’s claim that Claude solved longstanding black-hole physics problems, alongside work on model oversight, coding-agent efficiency and the stability of human-AI cooperation.