Welcome to this week's edition of Overclocked!

In this issue, big studios are testing copyright law against generative art while Apple bets a fresh feature can revive its AI buzz. We unpack both stories, scan six must-read headlines, and share Google’s latest prompt-engineering gems. ⬇️

In today’s newsletter ↓
😤 Disney and Universal are fed up with copyright infringement
👀 Apple’s Visual Intelligence hopes to wow
🏆 Google Chrome may finally have a challenger
🏀 AI-generated ads used during NBA finals
🎓 Weekly Challenge: Master Google’s prompt guide

🎬 Disney and Universal Sue MidJourney

On June 11, a joint complaint filed in California federal court pits Disney, Universal, and several other studios against image generator MidJourney for “industrial-scale” infringement. The suit claims MidJourney trained on “tens of millions” of copyrighted film stills, posters, and storyboards, producing outputs that sometimes replicate signature characters and scenes almost pixel-for-pixel.

📜 Studios Say Fair Use Falls Flat

Plaintiffs argue no transformative defense applies when users can prompt “Elsa in Arendelle, Pixar style” and receive near-identical imagery. They cite a MidJourney Discord gallery where fan prompts reproduced Spider-Verse artwork, alleging “consumer confusion and market substitution.” The filing demands statutory damages of up to $150,000 per infringed work and seeks a permanent injunction depriving MidJourney of Hollywood datasets.

🤖 The Generator’s Counterplay

MidJourney CEO David Holz told BBC the company will “vigorously defend artistic innovation” and may request partial dismissal, pointing to the U.S. Authors Guild v. Google Books ruling that deemed large-scale scanning permissible for search indexes. Legal scholars note, however, that MidJourney outputs—not just its model weights—contain recognizable IP, complicating fair-use arguments.

🌐 Wider Impact

If the studios win, AI labs could need licenses for entertainment content, raising barriers for smaller players. Creative pros worry a chilling effect might limit style transfer, while rights-holders cheer the chance to monetize data. The case also tests whether consent-based training agreements—similar to Shutterstock-OpenAI deals—become industry norm.

🗝️ Takeaway

This clash could redraw the line between inspiration and infringement for generative models. Courts must decide: is scraping public images innovation or appropriation? A ruling either way will echo far beyond Disney castles.

🍏 Apple Visual Intelligence Hopes to Wow 

Apple’s WWDC preview spotlighted Visual Intelligence, a new iOS 26 feature that turns any screenshot into a smart entry point for deeper actions. Tap-and-hold a captured clothing photo, and Visual Intelligence surfaces buy-now links; highlight a math problem, and it launches step-by-step solutions.

🖼️ Context Detects Everything

Under the hood, Apple integrates a lightweight vision transformer paired with on-device OCR. Unlike Google Lens, processing happens locally, preserving privacy. In early hands-on tests, CNET found the system correctly recognized obscure flowers and vintage vinyl covers in less than a second.

⚡ Why Apple Thinks It Matters

Cupertino’s broader Apple Intelligence push drew lukewarm reactions for lacking a headline capability. Visual Intelligence could change that narrative by delivering a tangible utility “regular users” feel daily. Macworld argues it bridges shortcuts and Spotlight, making screenshots the new command-line for iOS workflows.

🛑 Limitations and Privacy Lens

Recognition still stumbles on handwritten notes and comic fonts. Apple says all inference stays on Apple Silicon, but cloud look-ups trigger when users ask for shopping links—raising potential tracking questions. Developers await APIs promised later this year; if Apple grants hook-ins, apps could auto-tag bills, recipes, or even game walkthroughs.

Bottom Line: Visual Intelligence may not be ChatGPT in your pocket, yet it shows Apple’s strength: polish a single pain point until mainstream users wonder how they lived without it.

The Weekly Scoop 🍦

🌐 Arc releases Dia, an AI-powered Chrome challenger

📝 Sam Altman’s singularity blog post sparks debate

📺 AI-generated ads hit U.S. primetime TV slots

📉 Publishers blame traffic plunge on Google AI overviews

⚙️ Apple unveils AI-powered Shortcuts app at WWDC

🔮 Nvidia CEO calls quantum era the next inflection point

🎯 Weekly Challenge: Prompt Like a Pro

Google’s new prompt-engineering whitepaper distills years of research into practical tactics. Here are a few key takeaways about the best prompting strategies in 2025. You can also check out the entire whitepaper here.

Don’t forget to try the examples below also!

Key takeaways to remember about prompting in the modern age:

👓 Give an example
If the first reply misses the mark, show the AI one sample answer; it quickly picks up the pattern.

🏢 Assign a role
Start with “You are a friendly travel agent…” or “Act as a pediatrician…” to steer tone and expertise.

🤔 Ask it to think step-by-step
Adding “Think it through in steps” often produces clearer logic and fewer mistakes.

☑️ Vote on the best answer
Have the AI generate two or three responses, then ask, “Which of these is the strongest and why?” You’ll usually get a sharper final take.

🙇 Request structured output
When you need lists, tables, or a specific format, say so—e.g., “List pros and cons in a table.” Clear instructions beat guesswork every time.

1. Chain of Thought: Ask Gemini 2.5 Pro to “explain photosynthesis to a 10-year-old.” Now prepend “Think step by step” and compare clarity.
   

2. Role Play: In Claude 4, start with “Act as an immigration lawyer” before your question about H-1B visas. Note specificity gains.
   

3. Self-Critique: In ChatGPT 4o, follow any answer with “Evaluate your solution’s weaknesses.” Observe iterative improvement.
   

Share which tactic boosted output the most. Master prompts, master AI.

That's all for this week's Overclocked! From Hollywood lawsuits to screenshot smarts and prompt-hacking tips, AI isn’t slowing down—and neither are we. Keep experimenting, keep questioning, and remember: the future won’t write itself (yet).

Zoe from Overclocked