AI News — Wednesday, May 20, 2026
Google's official I/O 2026 collection highlights major advancements in AI, developer tools, and product integrations across its ecosystem, signaling a significant strategic shift.
Google's radical redesign of its iconic search box, driven by new AI capabilities, marks a pivotal moment in how users will interact with information retrieval.
Google introduces advanced AI agents at I/O 2026, enabling users to perform complex, multi-step tasks and go beyond traditional keyword-based searches.
OpenAI announces new initiatives to enhance content provenance, aiming to build trust and transparency in an increasingly AI-generated media landscape.
Google's I/O 2026 presentations position the company as a formidable force in AI-powered design tools, intensifying competition in a rapidly evolving market.
Users can now interact with their Gmail inbox using natural language, leveraging AI to manage emails, draft responses, and organize communications more efficiently.
New research explores 'Code as Agent Harness,' proposing a novel method to control and integrate AI agents using code, offering enhanced flexibility and reliability.
Researchers introduce SkillsVote, a system designed for the comprehensive governance of AI agent skills, from initial collection and recommendation to their ongoing evolution.
LongLive-2.0 presents an advanced NVFP4 parallel infrastructure, significantly improving the efficiency and capability for generating extended video content.
A new benchmark, CHI-Bench, assesses the ability of AI agents to automate end-to-end, long-horizon, and policy-rich healthcare workflows, highlighting their potential in medical applications.
Google Workspace introduces a suite of new AI-driven capabilities designed to streamline tasks, foster creativity, and boost overall productivity for users.
Google's AI Mode is reshaping how people search in the U.S., providing more intuitive and comprehensive results by leveraging advanced artificial intelligence.
Lance introduces a unified framework for multimodal modeling that leverages multi-task synergy to improve performance across diverse data types and applications.
This paper outlines a roadmap and user guide for leveraging AI to automate various stages of the research process, promising to accelerate discovery and analysis.
New research focuses on auditing the safety of AI agent harnesses, crucial for developing secure and reliable autonomous AI systems in various applications.