Amazon wins NYT, AI voices whisper & more (May 30, 2025)

Happy Friday!

It’s been a week of tipping points. The New York Times, long resistant to AI, just signed its first licensing deal. Voice clones are officially passing the human test. And an AI researcher made it into a top academic journal. I think we’re all just about ready for the weekend.

But first, let’s dive in.

🧠 AI Enters the Mainstream

📜 New York Times Strikes Licensing Deal with Amazon

One of journalism's final AI holdouts just gave in as The New York Times signed its first-ever AI licensing deal with Amazon. The agreement gives Amazon access to NYT editorial content, NYT Cooking recipes, and sports coverage from The Athletic. This content will feed into Alexa and Amazon’s AI models. The NYT’s ongoing lawsuits against OpenAI and Microsoft make this collaboration with Amazon particularly telling.

🎤 Human or Not? Voice AI Gets Spooky Real

Resemble AI’s new open-source tool, Chatterbox, clones a voice with just 5 seconds of audio, beating ElevenLabs in listener preference. Meanwhile, Hume’s EVI 3 understands emotional nuance and changes its tone depending on your mood. Between stammers, whispers, and excitement, it’s nearly impossible to tell it apart from a real person.

🔬 Intology’s Zochi Becomes First AI to Pass Peer Review

Zochi, an AI agent by Intology, just got a research paper accepted at ACL 2025, a top-tier NLP conference. It handled every step, including lit review, methodology, writing, and results, with only human help on formatting. Its multi-turn jailbreak detection paper scored in the top 8.2% of submissions.

⚙️ Product Updates & AI Power Plays

📊 Perplexity Labs Supercharges AI Reports

Perplexity’s Labs update now allows agents to run for 10 minutes, generating full dashboards and trading strategies — a major leap beyond its usual 3-minute limit. It’s now closer to a true research analyst assistant.

🖼️ Black Forest Labs Unveils FLUX.1 Kontext

Two FLUX.1 Kontext models just dropped: Kontext Pro (faster edits) and Kontext Max (better quality) can edit images using text while maintaining character identity and visual consistency. A Playground platform also lets businesses experiment before integrating.

🧠 DeepSeek R1 “Minor” Update Hits Major Milestone

DeepSeek R1 just tied OpenAI's o3 for second on the Artificial Analysis leaderboard. The update improved its math, coding, and reasoning skills. A new distilled 8B variant runs on a single H100 GPU and beats Gemini Flash in key tasks.

🛠️ Tools of the Day

• Everlyn – Generate visuals with a vision agent
• SchedX – AI agent that handles lead calls and bookings
• Wondera – Write and produce songs via chat
• Clado – Search people for sales, hiring, or research

🧠 Prompt of the Day

Expert Growth Consultant
Use this prompt to deeply audit any company’s competitive landscape and prioritize growth levers with an Impact × Feasibility matrix.

⚡ Quick Hits

• Meta is opening retail stores for its VR headsets
• Factory launches AI “Droids” to build software autonomously
• Kuaishou's Kling 2.1 competes with Google Veo 3

🧩 TLDR

The AI wave is crashing through old strongholds, from NYT’s unexpected partnership with Amazon to an AI earning a peer-reviewed paper acceptance. Voice models are becoming scarily human, agentic tools are leveling up fast, and DeepSeek’s "minor" update just made a major dent. If you blinked, you might’ve missed AI going full mainstream this week.

Enjoy the weekend,
David