Latest AI (Artificial Intelligence) News

📅December 21, 2025 at 1:00 AM

Major AI developments this week: regulation and safety laws, new model releases and partnerships, big investments, compute expansion, and AI-driven science initiatives.

UK boosts AI research funding and redirects UKRI budget

The UK is refocusing parts of UKRI’s £12 billion budget toward AI and advanced technologies to drive innovation and economic growth, signalling major public investment in AI research and infrastructure. This funding shift is intended to support strategic sectors, increase commercialization, and strengthen UK leadership in AI development.

OpenAI hires George Osborne to lead global Stargate initiative

OpenAI appointed former UK Chancellor George Osborne to lead its global ‘Stargate’ initiative, underscoring stronger ties between AI firms and UK policymakers. The move reflects companies’ efforts to navigate policy and expand international engagement amid tightening regulation.

New York passes landmark AI safety bill for major firms

New York enacted a state-level AI safety law requiring large AI companies to publish safety plans and report incidents, pushing accountability even as federal policy debates continue. The law is part of broader US state efforts to regulate AI following gaps at the federal level.

OpenAI releases GPT-5.2 amid intensified model competition

OpenAI launched GPT-5.2 as part of an accelerated product sprint responding to competitive pressure from rival models, continuing rapid iteration of large foundation models. Reports note heightened concern about misuse, such as synthetic image generation enabling photorealistic fakes.

Google DeepMind partners with US DOE for AI-driven science (Genesis Mission)

Google DeepMind partnered with the US Department of Energy to provide Gemini-powered AI tools to DOE national labs under the Genesis Mission, aiming to accelerate discoveries in fusion, materials, genomics, and weather prediction. Early outcomes include drug-repurposing candidates and validated predictions, with wider model access planned for 2026.

Global AI investment and data‑center deals surge in 2025

A report indicates global AI-related data‑center deals topped nearly $61 billion in 2025 as companies secure compute capacity for intensive AI workloads. Investors continue pouring capital into AI startups and infrastructure, sustaining a large funding pipeline into the sector .

Mistral AI teams with Tesco for three-year commercial partnership

Tesco signed a three‑year AI partnership with Mistral AI to enhance customer experience and internal workflows by deploying commercial models across retail operations. The deal is an example of retailers adopting advanced AI models to optimize supply chains and personalization.

Concerns rise about AI-generated disinformation and fraud

Journalists and researchers reported that new image-generation tools make creating convincing fakes easier, and scammers are exploiting AI-generated images for fraud and refunds, raising consumer-protection challenges. Security experts warn of growing misuse as models become more capable and accessible.

Academic initiatives expand AI-ready datasets and digital twins

Purdue launched an initiative to curate AI-ready datasets across geoscience, agriculture, life sciences and more, enabling AI-driven digital twins and reproducible research on its cyberinfrastructure. The program aims to accelerate discovery by providing indexed, real‑time data for model development and trustworthy AI deployments.

Faster, smaller AI chips and photonic computing breakthroughs reported in 2025

Researchers announced ultra-compact AI chips and optical/laser-powered AI computing approaches that promise much lower energy use and higher speed for inference and specialized tasks . Such hardware advances could reduce energy bottlenecks and enable more efficient edge and scientific AI applications.

Calls for integrating physics and scientific theory into AI models

Experts argue progress in AI requires combining large-scale data with scientific principles—constraining models by physics and causality to improve trustworthiness for critical applications like healthcare and climate. This trend supports building 'digital twins' and theory-informed models for safer, interpretable AI systems.

Industry consolidation and product launches continue in AI tooling

AI developer tools and code-assistant companies are consolidating: Cursor reportedly acquired Graphite while open-source coding models and autonomous engineering agents continue narrowing the gap with proprietary offerings. The tooling market is rapidly evolving as firms race to ship productivity-enhancing agents and IDE integrations.

Back to News