🤓 Nowhere to Hide: Truth Now Travels Faster Than Spin (Courtesy of AI)

PLUS: AI music goes mainstream; What AI studios need to win; Sora’s exciting updates; Claude gets memory—and more

I have a friend who loves action movie but hates when trailers give away the only good scenes.

Twice in the last few months, he asked ChatGPT about big studio releases—major action movies with major marketing budgets—and got the same warning:

“You can probably skip this one. The main action sequence is already in the trailer.”

He didn’t need to scroll through reviews or Reddit threads.

Within three minutes and a few back and forths, AI gave him exactly what he needed:

Direct answers to the questions that mattered to him the most.

He skipped both movies.

Because he trusted ChatGPT’s answer more than the Rotten Tomato score or any carefully crafted marketing message.

This is the new normal

And most businesses aren’t ready for it.

Marketing Can’t Outrun the Truth Anymore

For decades, entertainment marketing has run on a simple playbook:

✅ Create the ideal positioning early ➡️ Get a strong trailer out.

✅ Land the right press.

✅ Get the right quotes.

✅ Push that Rotten Tomatoes score up.

✅ Build momentum with media dollars and social on the final stretch.

And if the film doesn’t deliver on its promise, hope the opening weekend audiences buy tickets before word-of-mouth spreads too far.

It worked because information was scattered. Finding trusted opinions and figuring out the tradeoffs took time and effort. Making sense of it all took even more.

That friction gave marketers time.

Time to shape perception, build buzz, get butts in seats.

That time is gone.

AI tools like ChatGPT, Gemini, Perplexity are now the first stop for millions when they’re deciding what to watch or buy.

And they don’t just surface information.

They investigate on behalf of the user.

They pull from every corner of the internet: major outlets, niche blogs, Reddit threads, fan forums, YouTube comments.

They synthesize everything.

They personalize the answer based on tons of context from chat history.

And they deliver it in seconds.

All that complexity collapses into one helpful answer.

Just “the truth”. Or at least, the version of truth the AI can piece together from everything it finds across the most credible sources on the web.

And that version doesn’t care about your marketing spin.

What People Really Ask AI

Nobody’s asking ChatGPT: “Is the new Marvel movie any good?”

They’re asking:

“Will I like this if I’m not into superhero movies but loved The Batman?”

“Can I watch this with my 11-year-old without it getting too intense?”

“Is this show actually good, or is it just getting buzz because of the cast?”

“Is this laptop worth the upgrade if I mostly use it for video editing?”

“Does this restaurant live up to the hype, or is it just Instagram-pretty?”

These are specific. decision-making questions.

And AI treats them like research assignments.

Which means: You have less control over your brand’s narrative.

👉 Your carefully constructed story is now just one voice in a very loud room. AI breaks it apart and reassembles it, based on each person’s questions, preferences, and conversation history.

👉 That Rotten Tomatoes score you worked to optimize matters less now. People know scores can be inflated, and they want answers specific to what they care about.

👉 Your weaknesses are more visible. That “feel-good comedy” the trailer promised that’s actually kind of a downer? It’s showing up in AI summaries—because six people on Reddit brought it up, and the AI decided it was relevant.

Now apply that to your product launch. Your new services. Your public promises.

(While I’m mostly using movies and TV shows as examples, this applies across every industry—products, services, experiences, all of it.)

If there’s a gap between your story and reality, AI exposes it fast.

There’s no place to hide anymore.

Why This Is Actually Good

For audiences and f consumers, this shift is the best thing that could happen. They now have:

• An honest advisor who works only for them

• Less noise, more signal

• Smarter personalized recommendations in seconds

And it gives strong brands a real edge:

• Quality is harder to fake…and harder to miss.

• Overhyped and mediocre stuff die faster.

• Word-of-mouth scales more quickly when reality matches the pitch.

• Niche and unconventional projects find their right audience sooner.

The system rewards substance and match quality, and exposes brands that relied on friction, confusion, hype, or sheer visibility.

In a world where AI does the digging for you, the product has to stand on its own.

The Truth Isn’t a Liability—If You Frame It Right

The only sustainable strategy now is aligning your message so closely with the actual experience that when people ask AI about it, the answer matches what you already told them.

That means transparent communication that sets expectations, builds trust, and attracts the right audience.

If your show takes three episodes to find its footing, own that.

Frame the setup as an emotional on-ramp: the kind of investment that makes later payoffs hit harder. Make that part of the message.

Because when someone asks AI, “Is the first season worth pushing through if episode 1 and 2 are slow?” and the answer is, “Yes—episodes 3 to 5 are where it hits its stride,” (hello, Boardwalk Empire season 1😉)—it turns what might seem like a flaw into a deliberate part of the experience. It feels earned.

If your product has a learning curve but delivers once you master it, say that upfront.

Because when someone asks, “Is this easy to use right away?” and AI responds with “There’s a learning curve, but users say it becomes indispensable after the first week”, they’re more likely to give it a real shot.

The New Reality

AI is becoming how we all decide what deserves our time and money.

So, the real questions are:

👉 How do we evolve our strategies for a world where:

• Information is personalized by default

• The truth travels faster than spin

• There’s less control over the narrative and more transparency built in

👉 And how do we make sure AI reflects the story we want it to spread?

These are the questions I work through with clients every day.

They’re complex, but they’re exactly where the new competitive edge lives.

📈 Work with Me

AI Advising and Consulting: Strategic guidance to build sustainable AI strategies and adapt to shifting audience behavior

AI Training: Practical training that helps teams use AI more strategically, effectively, and safely—so it becomes a trusted thought partner, not just a task assistant

What You Need to Know About AI This Week ⚡

Clickable links appear underlined in emails and in orange in the Substack app.

---

📢 OpenAI’s Sora adds character cameos and video stitching tools.

You can now create reusable AI characters, including ones you’ve dreamed up from your drawings and doodles, and bring them to life across scenes.

You can even upload videos from your camera roll, like ones of your pet or your favorite stuffed animal, and turn them into a character cameo.

Each character gets its own display name and handle, along with customizable permissions.

You can keep characters private, share them with mutual followers, or make them available to everyone on the platform. You can also combine multiple clips to make longer multi-scene videos.

OpenAI is launching the feature with a selection of starter characters, including Halloween-themed options like Frankenstein, Dracula, Witch, and Ghost.

💰 Monetization + “Sora Economy”

Bill Peebles (who leads Sora at OpenAI) says paid credits are rolling out for heavy use (beyond 30 free gens/day) and that they’re piloting monetization: rights-holders will be able to charge for cameos of beloved characters and people.

This offers IP holders new ways to make money while fans get the chance to engage more deeply with the characters and worlds they love.

It could also open the door to new kinds of brand integrations: sponsored characters that are featured for a limited time, tied to launches, campaigns, or seasonal moments. (That part’s not official, but it’s not hard to imagine where things might be headed.)

So if you’re in marketing, I hope you’re paying close attention.

And for a limited time, Sora is open to new users without an invite in the US, Canada, Japan, Korea, Thailand, Taiwan, and Vietnam.

Check out OpenAI’s FAQ page here for step-by-step instructions on how to create cameo characters.

Since this rolled out yesterday, I haven’t had time to design a new character yet—or decide what kind of chaos she should cause.

But I did use Sora to animate my illustration from 👆 (which isn’t a new feature, but I couldn’t resist).

Now I want to play with it properly.

There goes my weekend…

🎥 Want to see what else I make in Sora? Follow me here.

---

🎬 Wonder Studios, an AI-powered production startup backed by OpenAI and Google DeepMind execs, just raised $12M to scale original content and IP.

It’s part of a growing trend: VCs betting AI can power full-stack studios.

But to compete with Hollywood, these new AI studios have to solve for far more than the tech and building workflows.

They’ll need to attract talented storytellers who bring vision, taste, and cultural pull—and find ways to support them that feel genuinely valuable.

That gets trickier as the tools improve, production gets cheaper, and more creators gain access and expertise.

When execution is abundant and cheap, vision holds the leverage.

And without the ability to offer real distribution or marketing muscle, it’s tough to convince the creatives to hand over their IP, ideas, or control.

As the tech keeps getting easier to use, many creators won’t have to rely on AI studios to get things made. They’ll either do it themselves or hire a small team of AI-savvy artists to help.

The risk: becoming workflow vendors for hire for someone else’s value chain.

Especially as the workflow layer keeps getting more commoditized and less profitable.

That’s the real puzzle.

Some will hopefully crack it. I’m rooting for them.

It will certainly be an interesting space to watch.

Gabriel Petersson, an OpenAI researcher on Sora, put it like this:

---

🧠 Anthropic rolled out memory for Claude—bringing its capabilities in line with ChatGPT and Gemini.

It now remembers past conversations, lets you see, edit or delete specific memories, and even offers “memory spaces” to keep work and personal threads cleanly separated.

You can also import/export memories from ChatGPT or Gemini, making it easier to switch (or experiment) without starting from scratch.

Memory is becoming table stakes for premium AI tools, especially for users doing real, sustained work with these models.

But memory without message capacity is a frustrating mismatch.

I absolutely love Claude, but even on the $20 paid plan, I get far fewer messages than ChatGPT (where I never hit a cap).

And when I’m mid-project, hitting that cap forces me to switch tools in the middle of a flow, or have to come back hours later.

This has made it hard for me to adopt it as a primary assistant.

That may soon change. Anthropic now has big infrastructure deals in place with both Google and Amazon, giving it access to the power it needs to scale up by early next year.

If that translates into higher message caps, Claude finally becomes a viable tool for real work, without needing to jump to the $100 or $200/month tier.

---

---

🎵 AI music just got too big to ignore.

Suno, an AI music generation app, has reportedly hit $150M in annual revenue.

The company has seen explosive growth, even while fighting lawsuits from the major labels.

It’s a clear sign of real consumer demand, and a willingness to pay for accessible music creation tools.

And despite ongoing industry backlash, most users don’t seem too concerned about how these models were trained (copyrighted data).

Taken together, it also proves AI generated music is a legit business, and one the industry can’t afford to ignore.

The majors are paying attention. Universal and Warner are now reportedly in talks with several AI companies, trying to shape the terms before someone else does.

And this week, Universal made the first major move.

Universal Music Group just settled its lawsuit with AI-music startup Udio, effectively turning it into a new business model.

Under the deal, Udio will license Universal’s catalog to train its next AI system, while UMG artists and songwriters get paid not just when their music is used, but when the model creates something new from it.

Udio’s next version, launching in 2026, will run inside a closed, licensed platform where users can remix, mash up, or use approved vocals—but can’t download or share their creations off-platform.

It’s a settlement, not case law, but a clear signal of where music and AI negotiations are heading.

--

Meanwhile OpenAI is reportedly working its own generative music generation tool that allows users to create music from text and audio prompts.

---

📊 OpenAI is paying over 100 former investment bankers $150/hour to create new training data that reflects the grunt work done in PowerPoint and Excel by junior bankers.

Why? Because those workflows have never been published on the internet. If you want AI to learn how analysts work, you have to build examples from scratch.

The goal is to make ChatGPT more useful in finance by teaching it to handle high-volume, low-creativity tasks—and to give OpenAI an edge others can’t easily copy.

Meanwhile Anthropic expanded Claude for Financial Services with Claude for Excel (an add-on that creates, analyzes and modifies spreadsheets) and connectors to real-time market data like Moody’s, LSEG and Alera.

It’s only currently available in beta as a research preview through a waitlist for 1,000 Max, Team and Enterprise plan customers, but they hope to expand access soon.

---

🌀 The internet thought it was AI. It wasn’t.

The video was scripted, staged, and shot by a Chinese comedian named Tianran Mu—who studied AI-generated clips so closely, he learned how to recreate their telltale flaws.

He nailed the blank stares. The awkward body movements. And the nonsensical plot twists.

That level of precision—of observational creativity—is what made it go viral.

Not the tech.

The talent.

What’s wild is that Mu doesn’t even have a TikTok, Instagram, or X account. Just a Chinese social media profile.

Other creators discovered the video and reposted it, pushing it past 11 million views.

It’s a reminder that: creators are gonna create.

In case you missed last week’s edition, you can find it 👇:

That's all for this week. See you either in 2 weeks or next Friday (if my schedule allows.)

Thoughts, feedback and questions are always welcome and much appreciated. Shoot me a note at avi@joinsavvyavi.com.

Stay curious,

Avi

💙💙💙 P.S. A huge thank you to my paid subscribers and those of you who share this newsletter with curious friends and coworkers. It takes me about 20+ hours each week to research, curate, simplify the complex, and write this newsletter. So, your support means the world to me, as it helps me make this process sustainable (almost 😄).