US military AI order; NVIDIA Vera and Microsoft MAI debut

Trump issues national security directive ordering US military to adopt AI

On 5 June 2026, President Donald Trump signed a national security memorandum directing the US military and intelligence agencies to accelerate their adoption of artificial intelligence. The memo requires updated doctrine for autonomous weapons systems, bars AI from being used to "censor free speech, embed ideological bias, or conduct unlawful surveillance," and calls on leading AI developers to voluntarily submit frontier models for government cybersecurity testing before public release. It is the first directive formally assigning AI adoption targets across the full US national security apparatus — from the Department of Defense to signals intelligence agencies — while retaining civil liberties provisions that oversight advocates had urged.

NVIDIA launches Vera CPU and RTX Spark for the AI-agent era at Computex 2026

At Computex 2026 in Taipei, NVIDIA CEO Jensen Huang announced two new chips built specifically for AI-agent workloads. The Vera CPU, with 88 custom Olympus cores and up to 1.2 TB/s of memory bandwidth, completes agentic AI and reinforcement learning tasks 1.8 times faster than x86 processors; Dell, HPE, Lenovo, and Supermicro plan to ship Vera-based servers. The RTX Spark superchip — a Blackwell GPU paired with a Grace CPU via NVLink-C2C — will deliver up to 1 petaflop of on-device AI performance in laptops and compact desktops from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, expected this autumn.

Microsoft debuts seven in-house MAI models at Build 2026

At Build 2026, Microsoft unveiled seven models under its new MAI (Microsoft AI) family, signalling a strategic shift toward self-sufficiency in AI. The flagship, MAI-Thinking-1, carries 35 billion active parameters and a 256,000-token context window, was trained on commercially licensed data, and matched Claude Opus on SWE-Bench Pro coding benchmarks in Microsoft's internal testing. A faster companion model, MAI-Code-1-Flash (5 billion parameters), achieved 51% on SWE-Bench Pro for high-volume coding automation. The full MAI family — also covering image, voice, and transcription workloads — is available on Azure AI Foundry.

AI labs jointly call for mandatory synthetic DNA screening

OpenAI, Anthropic, Google DeepMind, and Microsoft AI have co-signed a letter urging governments to require all synthetic DNA and RNA synthesis orders to be screened against databases of hazardous genetic sequences. The letter warns that advanced generative models could lower the barrier for actors seeking to design dangerous biological agents, and calls on sequence-synthesis companies to adopt mandatory screening immediately without waiting for legislation. It is the first joint biosecurity statement signed by all four major US AI laboratories.

Colorado's comprehensive AI Act takes effect 30 June 2026

Colorado's Senate Bill 205 (SB 205) becomes enforceable on 30 June 2026, making Colorado the first US state with a comprehensive AI risk-management statute. Developers and deployers of high-risk AI systems — those used in hiring, healthcare triage, education, and credit decisions — must complete impact assessments, implement risk-management programs, and disclose AI involvement to affected individuals. The Colorado attorney general leads enforcement; several other states are watching SB 205 as a template for their own legislation.

OpenAI forms enterprise deployment subsidiary

OpenAI has established the OpenAI Deployment Company, a subsidiary focused on large-scale enterprise adoption of its AI tools, acquiring consulting firm Tomoro to seed its Forward Deployed Engineering team. Backers include Bain Capital, Goldman Sachs, SoftBank, McKinsey & Company, and Capgemini. OpenAI disclosed that enterprise revenue now accounts for more than 40 percent of its total revenue, marking a shift in the company's commercial centre of gravity from consumer subscriptions toward institutional clients.