AI News Roundup: Anthropic Wins Injunction Against Pentagon, OpenAI Kills Sora, White House AI Framework

Judge Blocks Pentagon’s Blacklisting of Anthropic, Citing First Amendment Retaliation

In a stinging 43-page ruling issued yesterday, U.S. District Judge Rita Lin granted Anthropic a preliminary injunction blocking the Trump administration from enforcing its “supply-chain risk” designation against the AI company. “Nothing in the governing statute supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the U.S. for expressing disagreement with the government,” Lin wrote.

The dispute traces back to a $200 million Pentagon contract signed last July. Negotiations broke down when Anthropic drew two red lines: its AI would not be used for mass surveillance of American citizens or for fully autonomous weapons. The DOD wanted unfettered access for “all lawful purposes.” In late February, Defense Secretary Pete Hegseth declared Anthropic a supply-chain risk — and within moments, the DOD signed a competing deal with OpenAI.

More than 30 employees from OpenAI and Google DeepMind filed statements supporting Anthropic in their personal capacities. Judge Lin has delayed implementation of the injunction for one week to allow the government to appeal.

OpenAI Shuts Down Sora, Ends $1B Disney Partnership

OpenAI pulled the plug on Sora on March 24, shuttering the standalone video-generation app, the developer API, and a planned $1 billion partnership with Disney. The reason? Economics. At its peak, Sora was burning an estimated $15 million per day in inference costs while generating just $2.1 million in total lifetime revenue from in-app purchases.

Downloads had cratered from 3.3 million in November 2025 to roughly 1.1 million by February 2026. OpenAI said it will redirect compute resources toward “world simulation for robotics.” The Sora 2 model itself remains available within ChatGPT for Plus and Pro subscribers, but the dedicated product is dead.

White House Unveils National AI Legislative Framework

On March 20, the White House released its long-awaited AI policy blueprint, calling on Congress to create a unified federal approach that would preempt the growing patchwork of state AI laws. The framework recommends no new federal rulemaking body, instead relying on existing sector-specific regulators with a “light-touch” approach designed to preserve American competitiveness.

Key pillars include child safety protections (parental controls, age assurance for AI services), regulatory sandboxes for AI development, protections against AI-generated deepfakes, and leaving copyright questions around AI training data to the courts. Democrats immediately pushed back, introducing the GUARDRAILS Act to block the executive order underpinning the framework and preserve states’ authority to regulate AI independently.

OpenAI’s Acquisition Spree: Astral and Promptfoo

OpenAI has already made six acquisitions in 2026 — nearly matching its total for all of 2025. The two biggest March deals: Astral, the company behind uv, Ruff, and ty (widely-used Python developer tools), announced March 19; and Promptfoo, an open-source AI security and testing platform. Astral’s team will join OpenAI’s Codex unit, while Promptfoo’s technology will be integrated into OpenAI Frontier, the company’s enterprise AI platform.

Yann LeCun’s AMI Labs Raises $1.03B — Europe’s Largest Seed Round

Turing Award winner Yann LeCun’s new startup, Advanced Machine Intelligence (AMI Labs), closed a $1.03 billion seed round at a $3.5 billion pre-money valuation — the largest seed round ever for a European company. AMI is building “world models” based on LeCun’s JEPA (Joint Embedding Predictive Architecture), which learns from reality rather than just language.

The round was backed by Bezos Expeditions, NVIDIA, Samsung, Eric Schmidt, Mark Cuban, and Tim Berners-Lee, among others. AMI will operate from Paris, New York, Montreal, and Singapore.

MCP Hits 97 Million Monthly Downloads

Anthropic’s Model Context Protocol (MCP) — the open standard for connecting AI agents to external tools — reached 97 million monthly SDK downloads in March, up from roughly 2 million at its November 2024 launch. That’s 4,750% growth in 16 months. The ecosystem now includes over 5,800 community and enterprise servers spanning databases, CRMs, cloud providers, and developer tools. MCP has emerged as the de facto standard for AI agent tool integration, with adoption from every major AI company.