AI News Roundup: Judge Slams Pentagon’s Anthropic Blacklist, Apple’s Gemini-Powered Siri Imminent, Record AI Funding

Federal Judge Questions Pentagon’s Blacklisting of Anthropic

In a dramatic San Francisco courtroom hearing on March 24, U.S. District Judge Rita Lin told the Department of Defense that its decision to designate Anthropic a “supply chain risk” “looks like an attempt to cripple” the AI company. The case has become the highest-profile clash between Silicon Valley and the Trump administration over AI governance.

The dispute erupted after Anthropic refused to allow unrestricted military use of its Claude models — specifically opposing fully autonomous weapons and mass surveillance of American citizens. In retaliation, the Pentagon blacklisted the company, a move that could force defense contractors like Amazon, Microsoft, and Palantir to certify they don’t use Claude in any military work.

A newly surfaced court filing revealed that just one week before President Trump publicly cut ties with Anthropic, a Pentagon Under Secretary had emailed CEO Dario Amodei saying the two sides were “very close” to agreement. Judge Lin called this timeline “troubling.” A ruling on the preliminary injunction is expected within days.

Apple’s Gemini-Powered Siri Upgrade Could Ship This Month

Apple’s long-awaited Siri overhaul is finally on the verge of release. iOS 26.4, expected to ship by late March, will debut a fundamentally rebuilt Siri powered by Google’s Gemini models for complex reasoning and multi-step planning. The upgrade transforms Siri from a command-and-response utility into a context-aware AI assistant.

Key new capabilities include on-screen awareness — Siri can read and act on content currently displayed on the user’s device, such as booking a restaurant shown in Safari or adding a flight confirmation from email to the calendar. The system also supports natural multi-turn conversations and can chain multiple actions from a single request. Apple retains full control over the user interface, data routing, and privacy enforcement, with Gemini handling the reasoning layer behind the scenes.

White House Unveils AI Legislative Framework

On March 20, the White House released a national policy framework for artificial intelligence containing seven guiding principles for Congress. The framework’s central thrust: preempt “unduly burdensome” state AI laws while channeling federal oversight through existing agencies like the FTC, FCC, and SEC rather than creating a new regulatory body.

The recommendations span child safety rules, standardized permitting for AI data centers, and the creation of regulatory sandboxes to spur testing and development. The administration explicitly rejected a one-size-fits-all approach, saying states should retain the ability to enforce general laws protecting children, preventing fraud, and safeguarding consumers — but not to impose AI-specific regulations that could fragment the market.

NVIDIA’s Vera Rubin Platform and the $1 Trillion Bet

At GTC 2026, Jensen Huang unveiled the NVIDIA Vera Rubin platform — seven new chips now in full production designed to power every phase of AI, from massive-scale pretraining to real-time agentic inference. The platform pairs the Vera CPU with the Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, Spectrum-6 Ethernet switch, and the newly integrated Groq 3 LPU.

Huang raised NVIDIA’s sales projection to $1 trillion through 2027, citing the surging economics of inference. The company also announced general availability of NVIDIA Dynamo 1.0, an inference software platform described as “an operating system for AI factories,” and the Nemotron Coalition to advance open AI models through shared expertise, data, and compute.

OpenAI’s GPT-5.4 Era: From Frontier to Mini and Nano

OpenAI’s March has been defined by the GPT-5.4 family. The flagship model, launched March 5, unifies reasoning, coding, and agentic workflows into a single frontier model with a 1-million-token context window — the largest OpenAI has ever offered. It’s available in standard, Thinking, and Pro variants, with a 33% reduction in factual errors compared to GPT-5.2.

On March 17, OpenAI followed up with GPT-5.4 mini and nano, bringing frontier-class capabilities to smaller, faster models designed for high-volume workloads. Meanwhile, the company has been on an acquisition tear — six deals in 2026 alone, including Astral (developer tools) and Promptfoo (AI testing), nearly matching its entire 2025 tally.

AI Funding Hits Record Highs in March

March 2026 has already produced more $100M+ AI funding rounds than any comparable period in venture history. Yann LeCun’s AMI Labs raised $1.03 billion — the largest European seed round ever — to build world models based on JEPA architecture, backed by Bezos, Nvidia, Samsung, and Temasek.

Robotics is booming: Mind Robotics ($500M), Rhoda AI ($450M), Sunday ($165M), and Oxa ($103M) collectively raised over $1.2 billion in a single week for AI-powered industrial, household, and logistics robots. On the revenue side, OpenAI has surpassed $25 billion in annualized revenue and is reportedly considering a public listing as soon as late 2026, while Anthropic approaches $19 billion.