In partnership with

The #1 AI Newsletter for Business Leaders

Join 400,000+ executives and professionals who trust The AI Report for daily, practical AI updates.

Built for business—not engineers—this newsletter delivers expert prompts, real-world use cases, and decision-ready insights.

No hype. No jargon. Just results.

Subscribe free—trusted by leaders

Good morning! Today is Wednesday, June 4, 2025.

We have some exciting AI news today: Meta has signed a groundbreaking 20-year nuclear energy deal to support its AI infrastructure, and Phonely's AI agents have achieved a remarkable 99.2% accuracy in customer interactions.

1. Meta Strikes 20-Year Nuclear Power Deal to Fuel Its AI Growth

Meta has signed a groundbreaking 20-year deal to buy power from an Illinois nuclear plant, aiming to secure clean energy for its growing AI infrastructure. Partnering with Constellation Energy, the agreement will help cover the cost of upgrades and relicensing for the Clinton Clean Energy Center. While Meta won't use the power directly, it will claim the clean energy attributes to offset emissions from its data centers elsewhere. As AI demands skyrocket, Meta is joining other tech giants turning to nuclear as a stable, low-carbon solution for powering the future.

2. Anthropic Cuts Off Windsurf’s Direct Access to Claude AI Models

Windsurf, the fast-growing AI coding startup reportedly being acquired by OpenAI, says Anthropic has pulled back its direct access to popular Claude AI models with little warning. This sudden move forces Windsurf to rely on third-party providers, potentially disrupting service for developers using Claude 3.5 and 3.7. While Anthropic says access is still possible via API keys, the loss of direct integration adds cost and complexity for users. The decision comes just weeks after Windsurf was excluded from the launch of Claude 4, raising questions about Anthropic’s priorities as it builds out its own coding tools.

3. Anthropic Launches Blog Written by Its Own AI, Claude

Anthropic has unveiled "Claude Explains," a new blog entirely written by its AI model Claude and reviewed by in-house experts. The blog focuses on technical topics and showcases how humans and AI can collaborate to produce educational content. Rather than publishing raw AI output, Anthropic uses Claude to draft posts that are then refined by subject matter experts with real-world examples and context. The company plans to expand the blog's topics to include creative writing, business strategy, and data analysis, positioning it as a live demo of AI augmenting expert work.

Unbiased Business Insights, Every Week

Get a weekly knowledge digest of business insights from 1440’s Business & Finance newsletter. Expect concise rundowns, context-rich visuals, and curated links to keep you ahead of the finance curve. Join 1440 for crisp explanations of business trends —no MBA required.

Sign Up Here

4. Fortnite Introduces AI-Powered NPCs That Think for Themselves

Epic Games is giving Fortnite creators the power to build AI-driven NPCs that go off-script and generate responses in real time. Revealed during the “State of Unreal” keynote, the new tool in the Unreal Editor for Fortnite lets developers customize voices, delivery styles, and personality traits of characters. The demo featured a persuasive AI character named Mr. Buttons, created with just 20 lines of text. While the tech is still limited to turn-based interactions, it marks a big step toward more dynamic in-game storytelling. AI-generated dialogue, like a recent viral Darth Vader moment, shows the potential and pitfalls of this bold new feature.

5. AI Phone Agents Just Got Real: Phonely Hits 99% Accuracy and Replaces 350 Human Workers

Phonely’s new AI voice agents are so fast and accurate, most callers can’t tell they’re not speaking to a real person. In partnership with chipmaker Groq and optimization platform Maitai, Phonely has slashed response times by 74% and pushed conversational accuracy to 99.2%, beating GPT-4o. This breakthrough solves the awkward delay that’s long plagued AI phone systems and is already transforming call centres, one major customer is replacing 350 human agents this month. It’s a major leap forward for AI-powered customer service, and a sign that seamless, human-like AI conversations are no longer science fiction.

6. Knock and Cloudflare Just Made It Easier to Build AI Agents That Ask for Human Help

Knock’s latest integration with Cloudflare’s Agents SDK shows how AI agents can be smarter and safer by including humans in the decision-making loop. In a new demo, Knock’s messaging tools are used to build an AI assistant that helps issue virtual cards—but pauses and asks for a real person’s approval before completing the task. By combining real-time chat, durable state storage, and flexible messaging across email, Slack, and SMS, this setup allows developers to create scalable, human-guided AI workflows without complex infrastructure. It’s a big leap forward for trust and accountability in AI agents.

Stay tuned for more updates, and have a fantastic day!

Cheers,
Zephyr