Hi Creatives! 👋 

This week’s updates remind us that AI is moving fast—but it’s not just about speed. From sharper reasoning in GPT-5 to cinematic polish in Midjourney’s new video mode, and even policy debates about how AI should (or shouldn’t) be taught in schools, the landscape is shifting under our feet.

The thread connecting it all? Control and intention. Whether you’re building visuals, drafting strategy, or shaping education, these tools aren’t just getting smarter—they’re putting more creative power back in your hands.

This Week’s Highlights:

• 🚀 GPT-5: Smarter, Sharper, and Split into Three

• 🎬 Midjourney’s New Video & Moodboard Upgrades

• 🖌️ Photoshop’s AI Update (v26.10)

• ⚖️Regulate What AI Does, Not How It’s Built

• 📚 Should AI Be Taught From K–12?

• 🤖From Fear to Possibility✨

• 🎬 Creative Feature

🚀 GPT-5: Smarter, Sharper, and Split into Three

ChatGPT-5 has been released for a few weeks now. With that being said.

GPT-5 brings fewer hallucinations, stronger reasoning, and more control. It can auto-switch between “fast” and “deep-thinking,” handle longer context, and chain tools more reliably—making it a dependable upgrade for research, planning, and workflows. For creatives, this means sharper briefs, more consistent outputs, and reduced costs.

Our Take:
GPT-5 is a reliable upgrade for structure, research, and operations. But for brand voice and storytelling, many still lean on GPT-4o or adjust prompts heavily. Think of GPT-5 as the serious planner in the room, while GPT-4o still brings the charm.

🎬 Midjourney’s New Video & Moodboard Upgrades

Big updates just dropped on Midjourney—perfect for anyone experimenting with AI visuals:

• HD Mode for Video → Pro & Mega users can now generate in native 720p, giving crisper motion and detail (though it costs ~3.2× more than SD).

• Standard Plan Access → HD is now open to all plans, with smarter batch size controls (--bs 1 or --bs 2) to save on credits.

• Better Thumbnails & Moderation → Last-frame previews and fewer blocked jobs streamline your workflow.

Midjourney First & End Frames (Quick Guide)

• Start Frame: Pick any image (generated or uploaded) to begin your video.

• End Frame: Choose a different image, or reuse the start for a seamless loop.

🔧 How to set:

• Web: Upload in Starting/Ending Frame sections, or check Loop.

• Discord: Use --end [image_URL] for an end frame, or --loop for loops.

You can also adjust motion (--motion low/high) and extend clips up to ~21s.

✨ Why it matters: These updates put more control in your hands—whether you’re chasing cinematic polish, managing budgets, or dialing in the exact mood.

👉 Read the full article here

🖌️ Photoshop’s AI Update (v26.10)

Adobe just dropped a new wave of AI features this August—here are the highlights worth noting:

• Generative Upscale (beta): Enlarge images up to 8MP while keeping them sharp and detailed—great for print or giving new life to low-res files.

• Smarter Remove Tool: A stronger Firefly model powers cleaner, more realistic background fills when removing objects.

• Generative AI Model Picker: Choose which Firefly model drives your Generative Fill/Expand, giving you more style control.

• Projects (beta): Create shared workspaces to manage assets and collaborate more smoothly inside Photoshop.

Regulate What AI Does, Not How It’s Built ⚖️🤖

Last week, I had the privilege of attending the Stanford HAI Congressional Boot Camp plenary sessions—a gathering of policymakers, researchers, and industry leaders exploring how to shape AI governance, competitiveness, and infrastructure.

Here are the insights that stuck with me:

• Focus oversight on outputs → regulate harmful uses & measurable outcomes, not training budgets.

• Scope by impact, not inputs → obligations tied to revenue reach (e.g. $100M ARR), though even that may be too low.

• Context matters → healthcare, finance, and defense need different evaluation standards for the same model.

• Open source can be strategic → releasing one generation behind expands access while protecting the frontier.

• Safe shipping → policy-to-engineering translation layers, context-aware benchmarks, and independent red teaming.

• U.S. positioning → lean into distributed innovation, strengthen STEM education, and deploy AI-powered learning tools.

• Infrastructure reality → AI’s growth depends on diversified clean energy and stronger partnerships for data centers.

A huge thank you to the Stanford HAI team and all the speakers for sharing their perspectives—it was an incredible learning experience.

👉 Read my full reflections here: Regulate what AI does, not how it’s developed

📚 Should AI Be Taught From K–12?

If AI is shaping every industry, why aren’t we teaching it to kids early on?

👉 Starting September 2025, China will make AI education mandatory for every student—even six-year-olds.

• Each child will receive at least 8 hours of AI instruction per year

• Younger kids: hands-on projects

• Older students: robotics, machine learning, real-world AI applications

Meanwhile, here in the U.S., AI is still treated as an add-on, not a core subject.

From a creative perspective, I see both sides:
✨ Early AI literacy could raise a generation of creators who see these tools as a natural extension of imagination.
⚖️ But the challenge is designing programs, training teachers, and ensuring arts and play aren’t pushed aside.

This also echoes what came up during last week’s Stanford HAI Congressional Boot Camp: education and policy are moving far slower than AI adoption.

If AI is already shaping how we work, create, and connect—isn’t it time we prepare kids to engage with it, not just consume it?

🔗 Read more here

🤖From Fear to Possibility✨

Since late 2022, I’ve seen how fear and uncertainty can surround AI—but also how education can shift that conversation. Over the past year, I’ve been building spaces like The AI Stuff Creators and Artists Should Know and C.A.S.H. Camp to help creatives adapt, not resist. From Moodelier to Wonder Studios, my journey has shown me that the future of AI in creativity isn’t built alone—it takes an ecosystem of creators, technologists, policymakers, and communities working together.

clairexue

8,057 followers

View more on Instagram

🎥 I shared a short reel capturing this journey—watch it here

🎬 Creative Feature

Filmmaker Leo Kadieff has been experimenting with Kling AI’s new frames feature—and the results speak for themselves. By copying the last frame of one clip into the next, he’s creating smooth, cinematic transitions that feel effortless. His tests show how powerful this update can be for anyone working with AI filmmaking tools.

👉 Check out his full post here

👾 Alien Cyber-Tech Meets AI Music

Paige Piskin has been exploring new territory with AI-driven music experiments, blending storytelling, sound design, and visuals into a track about a cybernetic Medusa. Using tools like ElevenLabs, Veo 3, and CapCut, she layered operatic hums, mechanical “snake” sounds, and custom enhancements to craft an eerie, otherworldly soundscape—now live on Spotify.

Her workflow shows how AI tools can turn imagination into full sensory experiences, even for creators stepping into music production for the first time.

👉 Check out her full post here

🔄 Storytelling in Loops with Google Flow

Henry Daubrez introduces Ouroboros, an experiment that shows the storytelling potential of Google Flow. From just one image, he built a looping narrative of a girl trapped in a Groundhog Day-style cycle—waking in an overgrown apartment, tidying it, only to find it reset by morning.

Using Imagen 3, Veo 2, Veo 3, and Photoshop, Henry layered transitions, day-to-night shifts, and in-context B-roll to craft a seamless, evolving loop. It’s a striking example of how AI tools can transform static images into cinematic stories.

👉 See Henry’s full post here

📺 From Passive Watching to Playable Media

Samir Chaudry breaks down why Amazon’s investment in Fable Studio’s Showrunner—dubbed the “Netflix of AI”—signals a new era of media. Showrunner generates full TV episodes from prompts, complete with voices, characters, and story arcs. Even more, its series Exit Valley is “playable,” letting viewers step directly into the story.

Samir’s take: we’re moving beyond passive watching into interactive, hyper-personal media—where prompting, shaping, and even playing inside the content itself is becoming the new norm.

👉 Read his full post here

💡 Industry Insight

This week’s updates hit from all angles—GPT-5 getting sharper with reasoning, Midjourney stepping up its video game, Photoshop sneaking in more Firefly tricks, and even big questions like “should kids learn AI in school?” and “how should we regulate what AI does?”

For me, it all circles back to one thing: these tools keep evolving, but it’s the way we use them that matters. Whether you’re polishing visuals, drafting smarter ideas, or thinking about how the next generation will grow up with AI—your voice and choices shape what this tech becomes.

Play around with the new features, test them in your flow, and don’t forget: the best part of AI isn’t the upgrade—it’s what you create with it.

That’s a wrap for this week!