Microsoft leads AI model review as Google DeepMind and Meta agree to national security testing

The US government has signed agreements with Microsoft, Google DeepMind, and xAI to review their AI models for national security risks before public release. The Center for AI Standards and Innovation (CAISI) will conduct pre-deployment evaluations and research to understand the capabilities and risks of these AI models. This move aims to ensure that AI technologies are developed and used responsibly.

CAISI has already completed over 40 evaluations of AI models, including some that have not been released. The agreements are part of a broader effort to address concerns about AI posing a threat to cybersecurity and national security. Tech giants such as Microsoft, Google DeepMind, and xAI have agreed to share their AI models with the US government for national security testing.

In other developments, Anthropic's Claude Mythos Preview has shown remarkable capabilities in cybersecurity defense but also vulnerabilities in potential attacks. The model has raised concerns about its potential use in cyber attacks. Meanwhile, major publishers have filed a lawsuit against Meta Platforms, accusing the tech giant of using millions of their books and journal articles without permission to train its artificial intelligence model, Llama.

NVIDIA has introduced the Vera Rubin platform, which employs extreme co-design across multiple specialized chips and software optimizations to overcome throughput-latency tradeoffs. The platform is designed to handle the complexity of agentic AI architectures, which feature hierarchical agents and sub-agents that manage large context windows, tool calls, and memory statefulness.

Microsoft Dynamics 365 Supply Chain Management has enabled a more responsive operating model by connecting data, decisions, and actions across the supply chain. The company has also partnered with various organizations to develop AI systems that can detect and prevent cyber threats and improve decision-making.

Key Takeaways

['The US government has signed agreements with Microsoft, Google DeepMind, and xAI to review their AI models for national security risks before public release.', 'CAISI has completed over 40 evaluations of AI models, including some that have not been released.', "Anthropic's Claude Mythos Preview has shown remarkable capabilities in cybersecurity defense but also vulnerabilities in potential attacks.", 'Major publishers have filed a lawsuit against Meta Platforms for using their books and journal articles without permission to train its AI model, Llama.', 'NVIDIA has introduced the Vera Rubin platform to handle the complexity of agentic AI architectures.', 'Microsoft Dynamics 365 Supply Chain Management has enabled a more responsive operating model by connecting data, decisions, and actions across the supply chain.', 'The Center for AI Standards and Innovation (CAISI) will conduct pre-deployment evaluations and research to understand the capabilities and risks of AI models.', 'Tech giants such as Microsoft, Google DeepMind, and xAI have agreed to share their AI models with the US government for national security testing.', 'The agreements aim to ensure that AI technologies are developed and used responsibly.', 'The US government is increasingly showing interest in AI technology and security.']

US government to review AI models for national security

The US government has signed agreements with Microsoft, xAI, and Google DeepMind to review their AI models for national security risks before public release. The Center for AI Standards and Innovation (CAISI) will conduct pre-deployment evaluations and research to understand the capabilities and risks of these AI models. This move aims to ensure that AI technologies are developed and used responsibly. CAISI has already completed over 40 evaluations of AI models. The agreements are part of a broader effort to address concerns about AI posing a threat to cybersecurity and national security.

CAISI signs agreements with Google DeepMind, Microsoft, and xAI

The Center for Artificial Intelligence Science and Innovation (CAISI) has signed agreements with Google DeepMind, Microsoft, and xAI to collaborate on pre-deployment evaluations and research related to AI development for national security applications. The agreements aim to advance the development of AI technologies that can support national security missions. CAISI will work closely with these leading AI labs to develop AI systems that can detect and prevent cyber threats and improve decision-making. The collaborations will focus on developing AI systems that can be used to support national security missions.

Tech giants open AI models to federal national security tests

Tech giants such as Microsoft, Google DeepMind, and xAI have agreed to share their AI models with the US government for national security testing. The Center for AI Standards and Innovation (CAISI) will conduct safety testing of new AI systems before they are released publicly. CAISI has already completed over 40 pre-deployment evaluations of AI models, including some that have not been released. The agreements aim to support information-sharing and drive voluntary product improvements.

Microsoft, Google, xAI agree to share AI models with White House

Microsoft, Google DeepMind, and xAI have agreed to share their AI models with the US government for pre-clearances and security reviews. The Center for AI Standards and Innovation (CAISI) will conduct evaluations to better understand the capabilities and risks of new AI tools. The move aims to ensure that AI technologies are developed and used responsibly. The agreements are part of a broader effort to address concerns about AI posing a threat to cybersecurity and national security.

US announces deals with tech firms for national security review of AI models

The US government has struck deals with Google DeepMind, Microsoft, and xAI to review their AI models for national security risks before public release. The Center for AI Standards and Innovation (CAISI) will conduct safety testing of new AI systems. The agreements aim to ensure that AI technologies are developed and used responsibly. CAISI has already completed over 40 evaluations of AI models.

Microsoft, xAI, Google to give US government early look at AI models

Microsoft, xAI, and Google DeepMind have inked deals with the Commerce Department to allow the government to vet their AI models for national security risks ahead of release. The Center for AI Standards and Innovation (CAISI) will conduct safety testing of new AI systems. The agreements aim to support information-sharing and drive voluntary product improvements.

US government to test AI models, expand oversight

The US government will test AI models from top firms before release to vet them for security risks. The Center for AI Standards and Innovation (CAISI) has deals with Microsoft, xAI, and Google DeepMind. CAISI has completed over 40 evaluations of AI models so far. The government is increasingly showing interest in AI technology and security.

Google, Microsoft, and xAI agree to let US government test AI models

Google DeepMind, Microsoft, and xAI have agreed to let the US government test their AI models before public release. The Center for AI Standards and Innovation (CAISI) operates within NIST and was originally established in 2023 under Biden as the AI Safety Institute. The Trump administration renamed it last June.

Building for the Rising Complexity of Agentic Systems

Agentic AI architectures feature hierarchical agents and sub-agents that manage large context windows, tool calls, and memory statefulness. Real-world agentic sessions demonstrate token volumes scaling from tens of thousands to over 150,000 tokens per context window. The NVIDIA Vera Rubin platform employs extreme co-design across multiple specialized chips and software optimizations to overcome throughput-latency tradeoffs.

How to Build In-Vehicle AI Agents with NVIDIA

The automotive cockpit is undergoing a fundamental shift from rule-based interfaces to agentic, multimodal AI systems capable of reasoning, planning, and acting. Large language models, vision-language models, and speech models enable a fundamentally new interaction paradigm. NVIDIA DRIVE AGX platforms are ideally suited to meeting the requirements of in-vehicle AI.

From intelligence to impact

Agentic AI redefines the supply chain workflow by enabling proactive risk management, faster execution and orchestration, and coordinated human and agent workflows. The result is a more responsive operating model. Microsoft Dynamics 365 Supply Chain Management enables this model by connecting data, decisions, and actions across the supply chain.

Mythos AI is a cybersecurity threat

Anthropic's Claude Mythos Preview shows remarkable capabilities in cybersecurity defense but also vulnerabilities in potential attacks. The model has raised concerns about its potential use in cyber attacks.

AiTradeBtc Releases Comparative Overview of AI Crypto Trading Platforms

AiTradeBtc has released a comparative overview of AI crypto trading platforms, highlighting automation trends in 2026. The overview explores how AI trading systems process market data, execute strategies, and monitor risk dynamically.

Can Europe gain an edge on AI and tech

The POLITICO AI & Tech Week will discuss policy choices and tradeoffs happening in AI and technology in Europe. The event aims to make sense of the direction of travel in AI and technology.

Major publishers sue Meta for copyright infringement

Major publishers have filed a lawsuit against Meta Platforms in Manhattan federal court, accusing the tech giant of using millions of their books and journal articles without permission to train its artificial intelligence model, Llama.

Coinbase, PayPal Layoffs

Coinbase and PayPal are cutting staff, with AI cited as a factor. Experts discuss the shift towards profitability and efficiency driven by AI adoption in the tech sector.

He Couldn’t Land a Job Interview

A medical student, Alex, suspected that an AI-powered hiring tool was biased against applicants from underrepresented groups. He analyzed the hospital system's hiring data and found that the tool was indeed biased.

30 minutes saved per store

Kellanova has deployed AI-powered retail playbooks, resulting in 30 minutes saved per store and a 1.8% sales uplift. The company uses AI to analyze transaction-level data and guide investments.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

AI National Security Cybersecurity US Government Microsoft Google DeepMind xAI CAISI Pre-Deployment Evaluations AI Models Risk Assessment Responsible AI AI Development AI Technologies AI Safety NIST AI Adoption AI Bias AI-Powered Hiring Tool AI-Powered Retail Playbooks AI Crypto Trading Platforms Artificial Intelligence Machine Learning Agentic AI AI Architectures NVIDIA Vera Rubin Platform AI Agents In-Vehicle AI Supply Chain Management Cybersecurity Defense AI Trading Systems Market Data Risk Monitoring AI Adoption in Tech Sector Profitability Efficiency AI-Powered Hiring Tool Bias AI-Powered Retail Playbooks Benefits

Comments

Loading...