Greetings Qrew,

Whether you're diving into AI for work, side projects, or just out of curiosity — this issue of The Creators Hive brings you practical insights, hands-on learning, and tools you can actually use.

1. AI Avatars Outsell Humans in China’s $946B Livestream Market
2. OpenAI Signs $200M Pentagon Deal, Launches Government Division
3. Midjourney Finally Launches Its First Video Model
4. AI Watchdogs Release “OpenAI Files” to Expose Governance Concerns
5. MIT Study: ChatGPT May Hurt Student Cognition
6. Tool Recommendation of the Day: Augment AI (by Pollen)

---

AI Avatars Outsell Humans in China’s $946B Livestream Market

Chinese entrepreneur Luo Yonghao just livestreamed using his AI twin, raking in over $7 million in sales — outperforming his own human-led sessions and signaling a dramatic shift in live commerce.

Key Details:

• Two AI avatars hosted a 6-hour stream, promoting 133 products with human-like gestures and real-time viewer engagement.
• The broadcast pulled 13M viewers, beating Luo’s May human stream performance within just 26 minutes.
• Baidu’s ERNIE model generated over 97,000 characters of product copy for the session.
• This marks the first-ever livestream featuring dual digital hosts interacting naturally with each other and the audience.
• In China, over 100,000 digital humans are already working in e-commerce, cutting costs by 80% and boosting sales by 62% on average.

Why it matters:​
AI avatars are no longer just a novelty — they’re becoming the top salespeople in one of the world’s largest markets. As realism blurs the line between human and AI, businesses are prioritizing efficiency over authenticity — and customers may not even notice the difference.

---

OpenAI Signs $200M Pentagon Deal, Launches Government Division

OpenAI just unveiled "OpenAI for Government" — a unified initiative for its public sector work — while also securing a $200M defense contract with the U.S. Department of Defense to support national security, cyber defense, and military operations.

Key Details:

• The one-year DoD contract marks OpenAI’s first formal entry into U.S. military work.
• Projects include:
  • ChatGPT Enterprise to assist military personnel with tasks like navigating benefits.
  • Custom AI models for cybersecurity and “warfighting” prototypes.
• The initiative folds in existing collaborations with NASA, NIH, the Air Force, and the Treasury Department.
• Despite the military focus, OpenAI emphasized compliance with its use-case policies on safety and ethics.

Why it matters:​
The line between AI innovation and national defense is rapidly fading. As leading labs team up with governments and defense agencies, the global AI race is becoming a key front in geopolitical power and military modernization — raising serious questions about ethics, oversight, and the future of warfare.

---

Midjourney Finally Launches Its First Video Model

Midjourney just dropped V1, its long-awaited image-to-video model — turning stills into animated 5-second clips via the web, just as it faces major lawsuits from Disney and Universal over alleged copyright misuse.

Key Details:

• V1 lets users automatically animate or apply custom camera moves to any image (Midjourney or external).
• Each job creates 4 clips, extendable to 20 seconds, and costs 8x an image generation — still 25x cheaper than rivals, says Midjourney.
• Videos retain the platform’s signature aesthetic — a blend of surrealism and stylized realism.
• Currently web-only, with no audio support, keeping it distinct from full video generation rivals like Veo 3 or Sora.
• CEO David Holz called it an early step toward Midjourney’s long-term goal: real-time open-world simulation.

Why it matters:​
Midjourney’s cinematic touch finally hits motion. While it won’t rival the feature-rich, text-to-video giants yet, V1 marks a bold artistic entry into AI video, staying true to what makes the brand iconic — visual storytelling with unmistakable vibe.

---

AI Watchdogs Release “OpenAI Files” to Expose Governance Concerns

​Two watchdog groups — The Midas Project and Tech Oversight Project — just launched the OpenAI Files, a centralized archive of documents, testimonies, and analysis aimed at exposing transparency issues, governance breakdowns, and conflicts of interest at OpenAI.

Key Details:

• Covers four critical areas:
  Restructuring​
  CEO Integrity​
  Transparency & Safety​
  Conflicts of Interest
• Dissects OpenAI’s complex corporate structure and its transition to a Public Benefit Corporation (PBC).
• Includes a proposed “Vision for Change” — a framework urging OpenAI to meet the exceptionally high ethical standards needed for companies steering AGI.
• Provides evidence-based critiques in a transparent, public hub designed for both the public and policymakers.

Why it matters:​
As OpenAI moves closer to superintelligence, questions around who governs, how decisions are made, and whose interests are prioritized are intensifying. The OpenAI Files pull those questions out of the shadows — demanding accountability from the most powerful AI lab in the world.

---

MIT Study: ChatGPT May Hurt Student Cognition

MIT researchers found that students relying on ChatGPT for essay writing had significantly weaker brain activity and memory compared to those using traditional tools or writing without assistance.

Key Details:

• 54 students were split into 3 groups:
  No tools​
  Google search​
  ChatGPT
• Over 4 months, EEG scans tracked brain activity while writing SAT-style essays.
• ChatGPT users showed the lowest neural connectivity and scored worse on creativity, memory, and writing quality.
• Students who wrote without help had the strongest brain activity across all key cognitive functions.

Why it matters:​
The study suggests that early dependence on LLMs like ChatGPT may impair cognitive development — especially in areas like memory and critical thinking. As AI tools become more common in classrooms, educators may need to rethink how and when they’re used to avoid undermining learning itself.

---

Tool Recommendation of the Day: Augment AI (by Pollen)

👉 Website:https://www.augment.co​

What makes it unique?

• Augment isn’t just a chatbot or LLM — it’s like having a personal AI sidekick that watches how you work across apps and helps proactively.
• It learns from your workflows (e.g., email habits, file organization, task switching).
• Generates custom automation suggestions (e.g., “You always forward these reports — want me to automate that?”).
• Private by design — all data stays local/on-device or encrypted on your cloud.

Special twist:​
It builds custom GPT-like agents trained on your patterns without you writing a single line of code.

Use case magic:​
✅ Automate repetitive personal admin work.
✅ Have AI pre-draft your emails or Slack replies based on your style.
✅ Surface insights (e.g., “You spend 40% of your time on X — want help streamlining this?”).

Why it’s rare:​
Most AI tools wait for you to prompt them. Augment observes, predicts, and proposes helpful actions in real time.

---

Masterpieces from the Hive 🎨

Here’s what our community has been creating this week using the Hive Wall:

32 North Gould Street, Sheridan, Wyoming 82801
​Unsubscribe · Preferences​