AI News Roundup: White House AI Framework, NVIDIA GTC Highlights, Anthropic vs. Pentagon

White House Unveils National AI Legislative Framework

The Trump Administration released a sweeping AI policy framework on March 20, urging Congress to pass federal legislation that would preempt the growing patchwork of state-level AI regulations. The “light touch” blueprint lays out six guiding principles: protecting children online, safeguarding communities from data-center energy costs, respecting intellectual property, preventing censorship, enabling innovation, and developing an AI-ready workforce.

Crucially, the framework calls for no new federal rulemaking body — instead relying on existing sector-specific regulators. The White House wants the framework codified into law this year, but analysts note that bipartisan consensus will be difficult even with Republican control of Congress. The move signals the administration’s intent to keep the U.S. regulatory environment friendly to AI companies while clawing back authority from states like California and Colorado that have passed their own AI safety laws.

NVIDIA GTC 2026: Agentic AI and New Hardware Take Center Stage

NVIDIA’s annual GTC conference wrapped up last week in San Jose, drawing over 30,000 attendees from 190 countries. CEO Jensen Huang used his keynote to unveil the Vera Rubin platform, a rack-scale deployment system that combines compute, networking, and data processing for large AI data centers. The company also made the IGX Thor industrial-grade platform generally available for real-time physical AI at the edge.

On the software side, NVIDIA launched its Agent Toolkit and the open-source OpenShell runtime for building self-evolving AI agents. Major enterprise partners — including Adobe, Salesforce, SAP, Cisco, and ServiceNow — announced integrations. NVIDIA also debuted the Physical AI Data Factory Blueprint, an open reference architecture for generating and evaluating robotics training data, doubling down on embodied AI as a core strategic bet.

Anthropic’s Pentagon Lawsuit Draws Industry-Wide Support

The standoff between Anthropic and the U.S. Department of Defense continues to reverberate across the AI industry. After the Pentagon labeled Anthropic a “supply-chain risk” for refusing to allow its models to be used for mass surveillance or autonomous weapons, more than 30 employees from OpenAI and Google DeepMind signed a statement backing Anthropic’s position.

The dispute intensified when Anthropic CEO Dario Amodei publicly accused OpenAI of engaging in “safety theater,” calling Sam Altman’s statements about the defense deal “straight up lies.” Meanwhile, Google is quietly expanding its own Pentagon work, providing AI agents to the DOD’s 3-million-person workforce for unclassified tasks — growing its government user base faster than either rival.

Anthropic Launches Claude Marketplace

Anthropic unveiled its Claude Marketplace earlier this month, an Amazon-inspired platform where enterprise customers can purchase third-party software built on Claude. The initial catalog features products from Snowflake, GitLab, Harvey, Rogo, Replit, and Lovable Labs. In a notable move, Anthropic is taking zero commission at launch and allowing customers to apply unused Claude commitments toward partner subscriptions.

The marketplace represents one of the first attempts by a foundation model provider to evolve into a full platform company, extending its commercial footprint beyond model access into the broader enterprise software ecosystem.

Atlassian Cuts 1,600 Jobs to Fund AI Pivot

Australian software giant Atlassian laid off roughly 10% of its workforce — about 1,600 employees — to redirect resources toward AI development and enterprise sales. CEO Mike Cannon-Brookes framed the cuts as “self-funding” the company’s AI investment. More than 900 of the eliminated roles were in software R&D, with cuts affecting North America (40%), Australia (30%), and India (16%).

The restructuring follows a brutal year for Atlassian shares, which have lost more than half their value amid the broader software selloff. CTO Rajeev Rajan also stepped down. The layoffs join a growing pattern in 2026 of major tech companies citing AI as the rationale for workforce reductions, following similar moves at Block and others.

Robotics Funding Explodes in 2026

Humanoid robotics has emerged as 2026’s breakout investment category. In a single recent week, Mind Robotics ($500M), Rhoda AI ($450M), Sunday ($165M), and Oxa ($103M) collectively raised over $1.2 billion for AI-powered robots targeting industrial, household, and logistics applications. The sector is projected to draw more than $20 billion in funding this year.

Meanwhile, Fei-Fei Li’s World Labs closed a $1 billion round to advance spatial AI and 3D world models, with Autodesk investing $200M. And Yann LeCun’s AMI Labs raised $1.03 billion — the largest European seed round ever — to build world models based on JEPA architecture. The message from VCs is clear: capital is flowing from software copilots into physical AI systems.

EU Adjusts AI Act Timeline; UK Pauses AI Training Rules

The European Council agreed to streamline certain provisions of the AI Act, pushing back the compliance deadline for high-risk AI systems by up to 16 months until standards and tools are confirmed ready. The Council also added a new prohibition on AI-generated non-consensual sexual content and child sexual abuse material.

Across the Channel, the UK government reversed course on its plans to regulate AI training on copyrighted content, admitting the proposal was “rejected by most respondents to the consultation.” The u-turn leaves the UK without a clear framework for AI training data rights, even as the EU and US move ahead with their own approaches.