OpenAI has released a research preview of GPT-5, its next-generation language model, emphasizing not just increased capability but tighter controls around alignment, reasoning reliability, and misuse prevention.

Microsoft announced a major expansion of Copilot across Windows, embedding AI more deeply into the operating system itself. Rather than acting as a standalone assistant, Copilot now sits closer to system-level functions — helping manage files, applications, settings, workflows, and contextual tasks across the OS.

Courts in multiple jurisdictions have begun issuing early legal rulings involving artificial intelligence systems, addressing a question that has hovered over the field for years: who is legally responsible when AI causes harm?

Nvidia reported another surge in demand for its AI-focused chips, reinforcing its position as the critical infrastructure provider for modern artificial intelligence. According to Reuters, cloud providers, governments, and enterprises continue racing to secure Nvidia’s GPUs to power large language models, autonomous systems, and next-generation AI applications.

The European Union has officially begun enforcing key provisions of the AI Act, the world’s most comprehensive regulation governing artificial intelligence systems. After years of drafting, debate, and preparation, companies deploying AI in or affecting the EU must now comply with binding rules around transparency, risk classification, and accountability.

At the 2026 AI Impact Summit in New Delhi, delegates from at least 75 countries signed the Delhi Declaration, a voluntary framework outlining shared principles and goals for artificial intelligence development and oversight.

Google has unveiled Gemini 2.5 Computer Use, an AI model that can literally use a web browser — clicking, scrolling, and typing like a human. This means AI agents can now navigate websites, fill forms, and interact with digital interfaces in real time.

A chilling new report from the Washington Post highlights a growing AI biosecurity risk: machine learning models can now design new toxic proteins that existing screening systems fail to detect.

eta (Facebook & Instagram’s parent company) announced that starting December 16, 2025, it will begin using conversations users have with its AI tools to further personalize content and advertisements. (Reuters)

SoftBank has agreed to acquire ABB’s robotics division for $5.4 billion, signaling a bold push to merge robotics hardware with AI computation. (Reuters) The division includes factories, industrial automation arms, and robotics capabilities — assets that could be tightly integrated with AI to produce more adaptive, intelligent machines.