China’s AI Ambitions and the Pentagon’s New Strategy

In a rapidly evolving landscape of artificial intelligence, significant developments have emerged, particularly from China and the U.S. military, alongside advancements in AI models and applications.

Will AI fix prior authorization—or make it worse?

The U.S. government is testing an AI-driven program aimed at improving insurance coverage decisions, specifically focusing on the often cumbersome prior authorization process. This initiative raises questions about whether AI can streamline these procedures or potentially complicate them further, reflecting the ongoing debate about AI’s role in healthcare.

China’s new World Artificial Intelligence Cooperation Organization is President Xi’s clearest play yet for a parallel AI order

During the World AI Conference in Shanghai, President Xi Jinping introduced the World Artificial Intelligence Cooperation Organization, which aims to establish a framework for AI governance that operates independently of Western influence. This initiative includes the provision of 5,000 AI training slots for countries in the Global South and plans for cooperation centers with various international alliances, signaling China’s intent to create a new global AI order.

The Pentagon’s new AI playbook treats slow adoption as a bigger risk than imperfect alignment

The U.S. Department of the Navy has unveiled a new strategy that prioritizes rapid AI integration into military operations, emphasizing that slow adoption poses greater risks than the potential for misalignment. This strategy includes deploying large language models directly on naval vessels and establishing an AI war council to enhance decision-making in mission scenarios.

Open-weight models now match frontier cyber performance from just four months ago at a fraction of the cost

The British AI Security Institute has reported that open-weight models, such as GLM-5.2 and DeepSeek V4-Pro, have significantly closed the performance gap with closed frontier models in cybersecurity. This development, which occurred within just four months, raises concerns about the effectiveness of safety measures on these models, potentially leaving defenders ill-prepared for emerging threats.

Google Cloud’s Always-On Memory Agent Replaces RAG and Embeddings With Continuous LLM Consolidation on Gemini 3.1 Flash-Lite

Google Cloud has introduced the Always-On Memory Agent, a new implementation designed to enhance memory processing in AI applications without relying on traditional vector databases or embeddings. This innovative approach utilizes an orchestrator to manage continuous memory operations, allowing for more efficient data handling and retrieval in generative AI tasks.

Anthropic slashes Claude Fable 5 limits in Max and Team Premium and pushes Pro users toward API pricing

Anthropic has announced significant changes to its Claude Fable 5 offerings, reducing usage limits for Max and Team Premium plans while also adjusting pricing for Pro users to shift towards API rates. This decision reflects competitive pressures in the AI market and aims to maintain user engagement despite the limitations imposed.

Databricks hits $188B valuation, extending its run as AI’s favorite second act

Databricks has successfully repositioned itself as a leading AI company, achieving a valuation of $188 billion. This transformation has been complemented by research highlighting the cost efficiencies of open-weight AI models, reinforcing Databricks’ status in the rapidly growing AI sector.

Zyphra Releases ZUNA1.1: An Apache 2.0 EEG Foundation Model With Variable-Length Inputs From 0.5 To 30 Seconds

Zyphra has launched ZUNA1.1, an EEG foundation model that accommodates variable-length inputs, significantly enhancing its flexibility compared to the previous version. This new model, released under the Apache 2.0 license, demonstrates improved performance in reconstructing and denoising EEG data across a broader range of input durations.

Compiled automatically by the Tech AI Newsdesk from public AI-news sources and summarised in our own words.