🇺🇸 Trump Signs AI Executive Order: Voluntary Model Reviews, No Mandates
President Trump signed an AI executive order asking American AI companies to voluntarily submit frontier models for national security testing — a lighter-touch approach than the sweeping Anthropic government ban ordered hours earlier. The White House is asking firms to allow the National Institute of Standards and Technology (NIST) to evaluate models before public release for cybersecurity vulnerabilities. Axios reports that “AI companies breathed a sigh of relief” at the voluntary framework, which sidesteps the mandatory review requirements that industry had opposed. Meanwhile, the DC Circuit Court is exploring an off-ramp in the Trump administration’s legal standoff with Anthropic over the separate government-wide ban.
🏛️ Congress to Consider Restricting Military AI Use
Congress will weigh legislation to restrict how the military deploys AI systems, according to a report from The New York Times. The proposed framework would establish congressional oversight requirements for any AI system capable of autonomous targeting or lethal decision-making. The move comes as the Pentagon’s AI user base has reached 1.5 million personnel and the deployment of battlefield AI accelerates — with the Pentagon tapping Shield AI for low-cost drone swarm software under a new program tied to ongoing operations. Defense officials are urging caution, with some warning that battlefield automation is outpacing policy safeguards.
💰 Alphabet Plans $80B Fundraising for AI, Berkshire Invests $10B
Alphabet announced plans to raise $80 billion for AI infrastructure and data center investments, with Berkshire Hathaway committing $10 billion of that total. The massive equity raise underscores the staggering capital requirements of the AI buildout — Goldman Sachs estimates hyperscaler AI capex will reach $725 billion in 2026 alone. Google’s stock slid slightly on dilution concerns, but analysts note the raise positions Alphabet to compete with Microsoft’s $190B and Amazon’s $200B AI spending plans.
🤖 Microsoft Build: First Reasoning Model, Scout Personal Agent
Microsoft unveiled its first flagship AI reasoning model at Build 2026, signaling a major step toward independence from OpenAI ahead of the latter’s planned IPO. The model, designed to compete with Anthropic’s Claude Opus 4.x series, creates what analysts describe as a “thinking + programming” intelligent agent closed loop. Microsoft also launched Microsoft Scout, an always-on personal work agent powered by OpenClaw, and released seven new models total — including tools allowing developers to define AI agent behavior tests using natural language descriptions.
🔬 xAI Releases Grok Build 0.1 on API, Pentagon Deal Sealed
xAI launched Grok Build 0.1 as an agentic coding model now available via API, marking a significant upgrade to the Grok platform. In a separate development, Axios reports that Elon Musk’s xAI and the Pentagon have reached a deal to use Grok in classified military systems — a dramatic reversal given the legal controversies surrounding Grok’s image generation capabilities. The company faces a class action suit over alleged non-consensual deepfake generation, and the Pennsylvania Attorney General is leading a multi-state letter demanding changes to Grok’s content guardrails.
🇨🇳 Alibaba Launches Qwen3.7-Plus: Multimodal Agent Model with Vision
Alibaba’s Qwen team released Qwen3.7-Plus, a multimodal agent model adding vision, deep reasoning, tool invocation, and autonomous iteration capabilities on the Bailian platform. The model claims to outperform ChatGPT and Gemini in coding benchmarks, and is being integrated into Taobao for agentic shopping. Separately, Alibaba previewed Qwen-VLA, a unified Vision-Language-Action model for embodied AI and robot control.
🏢 Meta Slashes 8,000 Jobs in AI Restructuring
Meta laid off 8,000 employees (10% of its workforce) in a restructuring that accelerates the company’s pivot toward AI. The cuts follow Meta’s controversial shift from open-source Llama to the proprietary Muse Spark model, and reports that the Llama successor has been delayed. The layoffs are the largest at Meta since 2023, and come as Microsoft also announced buyout packages, highlighting a broader industry reallocation of resources toward AI infrastructure.
🏛️ Connecticut Signs AI Safety Law, Restricts Employer AI Use
Connecticut became the latest state to enact comprehensive AI regulation, as Governor Ned Lamont signed legislation establishing youth online safety protections, AI deployment rules, and workforce upskilling initiatives. The law notably restricts employer use of AI in hiring and termination decisions, requiring notice when AI contributes to layoffs (AI-caused RIFs) — a first-of-its-kind provision in U.S. state law.
💡 Mayo Clinic and Microsoft Partner on Frontier Healthcare AI
Mayo Clinic and Microsoft announced a collaboration to develop frontier AI models specifically for healthcare, aiming to enable earlier disease diagnosis and improved clinical decision support. The partnership will combine Mayo Clinic’s medical data assets with Microsoft’s AI infrastructure and model-building capabilities.
In Brief
- Nvidia at COMPUTEX/GTC Taipei: Jensen Huang announced RTX Spark, a new PC AI chip line aimed at competing with Apple, Intel, and Qualcomm, alongside the previously reported $40B in AI equity deals and $58.3B quarterly profit.
- Alphabet’s $80B raise is part of a broader trend — Goldman Sachs predicts $725B in hyperscaler AI capex for 2026, with SoftBank’s €75B French data center investment contributing to the global infrastructure sprint. VanEck launched the RACK ETF to capture the data center supply chain buildout.
- Paste launches MCP support: The popular clipboard manager added Model Context Protocol (MCP) support, enabling users to connect their clipboard history to AI tools — the latest MCP ecosystem expansion following the NSA’s security guidance.
⏰ Limited-Time Free Models
Several limited-time free model promotions are expiring soon:
| Model | Expires | Pricing | Context |
|---|---|---|---|
| Qwen3 30B A3B | June 5 | $0.09/M in / $0.45/M out | 131K |
| Llama 3 Euryale 70B v2.1 | June 5 | $1.48/M tokens | 8K |
| Hermes 2 Pro Llama-3 8B | June 5 | $0.14/M tokens | 8K |
| Claude Opus 4.6 Fast | June 29 | $30/M in / $150/M out | 1M |
Act fast on the June 5 expirations — the Qwen3 30B A3B offers exceptional value at 131K context for just $0.09/M input tokens.
🚀 New Models on OpenRouter
- MiniMax M3 — A new model with an impressive 1M context window, now available on OpenRouter at $0.3/M tokens input. MiniMax is positioning M3 as a long-context alternative for document analysis and multi-turn agent conversations.
🤗 Trending on HuggingFace
Notable uploads to the Hub include AesSedai/Step-3.7-Flash-GGUF (2,007 downloads, GGUF quant of StepFun’s 3.7 Flash), nphSi/Z-Image-Lora (170K downloads, popular LoRA for image generation), and OpenTransformer/AGILLM-4 — a 4-parameter research model focusing on long-context language modeling with 6 likes. The ecosystem saw a relatively quiet day with mostly experimental uploads.
⭐ GitHub Trending: AI Edition
| Repository | Stars | Language | Description |
|---|---|---|---|
| pewdiepie-archdaemon/odysseus | 32,142 | JavaScript | Self-hosted AI workspace — continues explosive growth |
| GordenSun/GordenPPTSkill | 1,549 | Python | AI-friendly PPT builder with 17 hand-polished Chinese templates |
| asz798838958/aBaiAutoplus | 972 | Python | Multi-platform AI account auto-registration tool |
| withkynam/vibecode-pro-max-kit | 720 | JavaScript | Spec-driven coding harness for AI-assisted development |
| 2aronS/Duel-Agents | 681 | TypeScript | CLI, SDK, and IDE plugins for multi-agent orchestration |
💡 Key Trends
- AI Policy Enters New Phase: From Trump’s voluntary EO to Congress debating military AI limits, the policy landscape is rapidly fragmenting — no single federal framework, but a patchwork of state laws (Illinois, Connecticut), executive actions, and congressional initiatives.
- Capital Supercycle Intensifies: Alphabet’s $80B raise, Meta’s 8,000 job cuts, and Microsoft’s model independence push all point to a market where only the best-capitalized players can compete at the frontier.
- From Chat to Agents to Swarms: Microsoft Scout, xAI Grok Build 0.1, and Alibaba Qwen3.7-Plus all converge on agentic AI as the dominant paradigm — models are no longer judged on chat quality alone but on their ability to reason, use tools, and act autonomously.