Mato just raised pre-seed Read more

iPhone 18 Pro & Ultra Leaks Audit | Pixel Audit

  • Aug 25, 2026
  • 17 min

Show notes

What the episode covers

iOS 27’s Liquid Glass lock screen looks flashy, but in this Tuesday episode for week 35 of 2026, Grant and Maya argue it’s one of Apple’s riskiest recent UI bets. This design critique episode of Pixel Audit walks through how their own Apple event predictions hold up before the keynote: the September 9th date call, the Dynamic Island shrinking rumors, a crease-free Foldable Ultra, and the refined Liquid Glass effect in the iOS 27 beta.

They unpack why solid UX and product design work means separating confirmed behavior from speculative leaks, and how to judge sources without getting swept up in hype. Along the way, they break down specific UI decisions and their impact on real interactions.

  • Event date pattern analysis vs. blind guesswork in prediction culture
  • The Dynamic Island’s potential shrink and its impact on Live Activity targets
  • The rumored Liquidmetal hinge and crease-free foldable Ultra hardware
  • The new Liquid Glass transparency slider and icon rendering tweaks

Listeners will walk away knowing how to audit predictions, question leaker confidence, and run a sharper design critique of any UX or product design rumor they see.

Timeline

In this episode

7 moments worth skipping to. The timecodes match the player above.

  1. 0:12Introduction
  2. 1:20Event Date Countdown: Grading the Call
  3. 4:27Dynamic Island: Shrink or Stay?
  4. 7:50Foldable Ultra: Killing the Crease
  5. 11:05Liquid Glass: Refinement or Retreat?
  6. 14:05Scorecard Wrap: Confidence vs. Confirmation
  7. 16:54Outro

Quick answers

Straight from the episode

The questions this one settles, without the listen.

How did UX Autopsy grade the iPhone event date prediction in this episode?
Grant and Maya give the September 9th event-date prediction a “green flag with an asterisk.” They credit 9to5Mac’s four-year pattern analysis—independently echoed by MacRumors and Forbes—but stress that only the reasoning gets graded now, and the result isn’t truly confirmed until Apple actually ships.
Why did the hosts leave the Dynamic Island ‘shrink or stay’ prediction ungraded?
Because the leaks conflict and the prediction itself was too fuzzy. MacRumors cited Gurman and Weibo leakers saying the Dynamic Island would shrink, while Digital Chat Station pushed that slimmer cutout to iPhone 19. Grant and Maya admit they hedged between both timelines and argue that “smaller” without concrete interaction details isn’t a real, testable prediction yet.
What’s behind the ‘crease-free’ Apple foldable Ultra rumor discussed in the episode?
The crease-free claim comes from Yanko Design’s August 21st report about a Liquidmetal hinge with a teardrop-shaped cavity. The catch is that no outlet has independently verified that mechanism, so Grant and Maya treat the no-crease prediction as ungradable before the keynote and highlight how scoop-driven incentives bury those caveats.
How did the episode evaluate Apple’s Liquid Glass changes in iOS 27 beta?
They found the Liquid Glass refinement actually testable: Tom’s Guide reports a new transparency slider and tweaked rendering that fixes blurry, tilted icons and mismatched keyboard styling, while keeping the core effect. Grant and Maya mark this as a “green flag with a footnote,” pointing to the slider as a smart response to the original iOS 26 Reddit backlash.
What is the final prediction scorecard the hosts land on for this Apple cycle?
Out of five major calls, only two are currently gradable: the event date and the Liquid Glass refinement. Three remain unconfirmed before the keynote—the Dynamic Island shrink, the crease-free hinge, and Yanko Design’s $2,000 price plus September 18th launch for the foldable Ultra—so they’re all labeled ungraded rather than right or wrong.
What’s the main takeaway about handling Apple leaks and predictions from this episode?
Separate confidence from confirmation. Grant and Maya argue that “unconfirmed” isn’t an insult—it’s just honest labeling—and suggest using leaks to audit sourcing quality, not to pre-judge Apple’s real decisions. The listener takeaway: know what’s actually verified versus what you’re just hoping a leaker nailed.

Transcript

The full conversation

Every word of the episode, 2,848 of them, in the order they were said.

Read the transcriptHide the transcript

GrantOkay, so I pulled up our doc from last week, and Maya, we made some bold calls in here.

MayaBold is one word for it.

GrantThis is Pixelated. I'm Grant, that's Maya, and today we're doing something we've never actually done before the show.

MayaWe're grading ourselves before Apple's even said a word.

GrantRight. The keynote's not until September ninth. Nothing's confirmed yet.

MayaSo this isn't a victory lap.

GrantNo, it's more like checking your homework before the teacher's graded it.

MayaSome of these predictions we can basically confirm right now just from how solid the leaks are.

GrantAnd some of them we can't touch honestly until Apple's actually on stage.

MayaThe Dynamic Island call, the foldable crease thing, those are still just rumors dressed up as certainty.

GrantWhich is the real test here, right? Can we tell the difference between a good source and a guess we want it to be true?

MayaExactly. So let's find out where we actually stand.

GrantDeal. No cherry-picking the wins, no burying the misses.

MayaWhole scorecard out loud.

GrantStart with the easy one.

MayaThe date itself.

GrantMm-hmm.

MayaOkay, so the site 9to5Mac ran the numbers on this. Four years running, Apple drops the invite exactly 14 days before the keynote.

GrantEvery single year?

MayaFour for four. That's not luck. That's a pattern you can set your watch by.

GrantOkay, I'll give them that. That's a real methodology, not a vibe.

MayaRight? And once you back it out from the keynote date, September ninth lands exactly where the math says it should.

GrantSo they weren't guessing a date out of a thin air, they were reverse engineering it from Apple's own habits.

MayaWhich is honestly the only way I trust a prediction like this.

GrantMeanwhile, MacRumors and Forbes are both out there in late August with their own scheduling pieces just kind of triangulating the same window.

MayaNobody had the invite yet.

GrantNobody had the invite. They're all working off the same historical breadcrumbs, just publishing on slightly different days.

MayaOne's the nineteenth, one's the twenty-fourth.

GrantRight, and neither of them breaks new ground. They're just confirming the shape everybody already suspected.

MayaSo for the actual grade, I'm giving this a green flag.

GrantWhoa, hold on. Easy.

MayaWhat? The date hit.

GrantThe date hit, sure, but green flag implies the reasoning was airtight going in, and it wasn't airtight. It was just the best bet on the table.

MayaOkay, Grant, come on. A pattern that holds four years straight is airtight enough for me.

GrantIt's strong. I'm not saying it's weak. I'm saying strong pattern matching and a confirmed date are two different things, and we only get to call it confirmed because the keynote actually happened.

MayaOkay, fair, fair. Before Apple's invite dropped, this was still a bet.

GrantA really good bet. I'd have made the same one.

MayaSo what do you want, an asterisk?

GrantKinda, yeah. Green flag with an asterisk that says, "Correct because the pattern held, not because anyone actually knew."

MayaI can live with that asterisk.

GrantThat's the standard we're using all episode. Solid reasoning gets credit. Results only get confirmed once Apple actually ships them.

MayaDeal. Consistent rules for every call.

GrantGood, because that's basically our whole scorecard problem in one call. Some of these we can grade clean. Some of them we can't grade at all yet.

MayaAnd the date was the easy one.

GrantSure, pass?

MayaSure, pass.

GrantThe date was the easy one because it was already public by the time we were recording this.

MayaRight, which is not true of the harder stuff.

GrantNot even close. The second we get into what Apple screens actually look like, we're back to leak territory.

MayaUgh, which brings us to the Dynamic Island.

GrantYeah. That one we genuinely don't know. Yeah, so here's the file MacRumors ran back in May.

MayaThe Dynamic Island CAD leak?

GrantThat one. MacRumors says Bloomberg's Mark Gurman plus a pile of Weibo leakers were all pointing the same direction. The island shrinks, but it doesn't vanish.

MayaWhich is basically what we bet, right? We didn't say gone. We said smaller.

GrantWe said smaller, but smaller how? That's the part I don't think we actually nailed down.

MayaOkay, wait. Walk me through what actually changes if the housing shrinks.

GrantThe sensor stack has to get tighter. Face ID hardware, the camera, all of it crammed into less real estate. That usually means Apple's engineers found a way to consolidate components, not remove them.

MayaSo the pill gets narrower.

GrantProbably, but here's where our prediction gets messy.

MayaDigital Chat Station.

GrantExactly. Same MacRumors piece. Digital Chat Station is the leaker saying, "No, the slimmer cutout isn't an 18 thing at all. It's an iPhone 19 thing."

MayaSo one camp says it shrinks now, one camp says it shrinks later.

GrantRight, and we picked the middle.

MayaWe picked both.

GrantWe picked both. Go back and listen to what we actually said last time. It was, quote, "It'll probably be smaller unless it isn't."

MayaThat's not a prediction. That's a shrug with confidence.

GrantIt's a hedge dressed in analysis. If Digital Chat Station's right and nothing changes until next year, we don't get credit for guessing smaller.

MayaBut if Gurman's right, we do?

GrantWe get partial credit because we didn't say when, and when actually matters here. That's the whole design story. A smaller cutout changes touch targets on the Face ID unlock animation. It changes how the pill expands for Live Activities.

MayaIf the housing's tighter, does the animation expand even have room to breathe the same way?

GrantThat's my worry. If they shrink the physical cutout but keep the same interaction states, something's getting cropped, or the animation timing shifts to compensate.

MayaWalk me through the worst case. Say the pill's narrower, but the interaction code hasn't changed. What actually breaks for someone using it? The easiest failure is the Face ID unlock lift getting clipped mid-animation, so it looks like it's stuttering instead of resolving smoothly. Same risk on a Live Activity expanding. If the target area shrinks but the touch zone doesn't get remapped, people start missing taps on things like timer controls.

GrantThat's not a cosmetic problem. That's a functional one.

MayaRight. And neither leak actually tells us if Apple solved that part. Gurman's reporting the hardware direction, not the interaction spec.

GrantSo we can't grade this until we see the actual unlock flow on stage.

MayaWe can't. Not honestly.

GrantOkay, so on the scorecard, this one's sitting where?

MayaRight now it's ungraded. Two leakers, two timelines, and we split the difference instead of committing.

GrantRight. Harsh, but noted.

MayaIf that's what a hedge looks like on hardware we can literally see in photos, imagine what happens once we get to something Apple hasn't shown anyone yet.

GrantYou mean the fold.

MayaI mean the fold.

GrantThe whole pitch for the Ultra was that Liquid Metal hides the crease.

MayaAnd we went on record saying that.

GrantWe went hard on record. So let's see if that one holds up any better than this did. So the fold. Yanko Design ran a piece back on August twenty-first laying out what's supposed to make the crease disappear.

MayaRight. The hinge thing.

GrantLiquidmetal Technologies hinge with this teardrop shaped cavity built to spread out the bend stress instead of concentrating it at one seam.

MayaOkay, that actually sounds like real engineering, not just marketing copy.

GrantIt sounds like real engineering because somebody described it in engineering language. That's different from somebody testing it.

MayaFair. So what does the report actually claim happens?

GrantCrease free. That's the word they used. Not reduced, not minimized. Gone.

MayaAnd we bet on that on air fully.

GrantWe did. And here's the part that should have stopped us. The same piece admits nobody's verified the mechanism. No publication has actually put hands on a unit and confirmed the cavity does what it's supposed to.

MayaWait, so the outlet reporting the fix also says the fix is unconfirmed?

GrantIn the same breath, basically.

MayaThat's a little wild.

GrantIt's not malicious. It's just incentive. A confident sounding scoop gets more clicks than a hedged one, so the hedge gets buried in paragraph nine.

MayaWhich means readers and apparently us skim right past the caveat.

GrantIt's not wild. It's just how the leak cycle works. Somebody gets a hinge diagram, somebody else gets a spec sheet, and the write up stitches it into a story that reads like a finished product.

MayaBut it's a rendering of a rumor.

GrantIt's a rendering of a rumor with really specific vocabulary. Teardrop cavity, stress distribution. That stuff makes you trust it more than you should.

MayaOkay, so where does that leave our prediction? Because I remember being pretty confident recording that one.

GrantYou were. I was too. And I think that's the actual problem with this whole rewatch.

MayaGo on.

GrantThe island call, we at least had photos and two leakers disagreeing on timing. This one we don't even have that. We have one outlet's synthesis of a claim nobody's tested.

MayaSo it's not gradable, not pass, not fail.

GrantIt's not gradable yet, and I don't think we can pretend otherwise just because we want a clean scorecard today.

MayaWhich kind of blows up the format we planned, doesn't it?

GrantA little. We came in thinking we'd flip through and stamp each one.

MayaAnd instead we're finding out half of these were never fully checkable before the keynote happened.

GrantWhich honestly tells you something about how these predictions get made in the first place. You're not grading facts. You're grading how much you trusted the sourcing under pressure.

MayaThat's a different show than the one we planned.

GrantIt's a better one, I think. Harder, but better.

MayaOkay. So hardware's stuck in limbo. Island's a maybe. Crease is a maybe.

GrantBoth stuck the same way. Rumor shaped, not confirmed.

MayaWhat's left that we can actually check without waiting on a keynote?

GrantSoftware. Whatever's running on top of all this glass and metal, we can put our hands on that ourselves.

MayaLiquid Glass.

GrantLiquid Glass. Let's see if that holds up better than the hardware bits did. Okay. Liquid Glass. This one we can actually touch.

MayaFinally. No leakers, no anonymous tipsters. Just us and the beta.

GrantTom's Guide put the iOS twenty-seven beta side by side with iOS twenty-six, and Apple added a transparency slider.

MayaWait, a slider? Like you can just dial down the glass effect yourself?

GrantThat's what they're describing. And they say Apple also adjusted the rendering itself, trying to fix the blurry look.

MayaThe tilted icons thing.

GrantRight. People were saying the icons looked tilted, kind of warped depending on the wallpaper behind them.

MayaAnd the keyboard was inconsistent too, remember? Sometimes it looked frosted, sometimes it almost looked solid.

GrantSame report. They're calling that out as one of the things Apple tuned.

MayaOkay, so this is Apple actually listening.

GrantTo a point. Toms Guide is clear the underlying effect is still there. They tuned it. They didn't rip it out.

MayaWhich honestly makes sense. Remember when this whole thing first shipped?

GrantThe complaints piled up fast.

MayaThere was a Reddit thread. Someone said Apple went for looks at the expense of usability and functionality. Blunt, but people agreed.

GrantAnd in that same conversation, you had people defending the lock screen redesign, saying it actually looked great.

MayaSo it was never one verdict. Some people hated the transparency. Some people loved the new look.

GrantWhich is exactly why a slider is the smart move. You don't have to pick a side for your entire user base.

MayaOkay, quick fire. Green flag or red flag on the slider?

GrantYou go first. You're dying to.

MayaGreen flag. Big one. They took real feedback, people specifically complaining about legibility, and gave users a dial instead of a lecture.

GrantI'll play the other side for a second.

MayaHere we go.

GrantThe effect Tom's Guide is describing, the one causing the tilted icons in the first place, that's still baked into the OS. They tuned the intensity. They didn't rethink whether translucent everything was the right call for a phone interface.

MayaSo you're saying it's cosmetic?

GrantI'm saying it's a good patch on a decision they're not walking back. Big difference.

MayaFair, but from a user seat, if the icons stop looking warped and the keyboard stops flip-flopping, I don't care if the philosophy changed.

GrantNo, that's fair too. I'll take green flag with a note attached.

MayaA note, not an asterisk. You've used that one already today.

GrantCaught. Fine, a footnote.

MayaWhich is honestly a lesson bigger than Apple. Give people the dial before they revolt in a Reddit thread.

GrantFine. Fine. The point stands. A slider is a design admission that one setting can't serve everybody's eyes or wallpapers.

MayaSo out of everything on our list, this is the one where we actually have hands-on beta behavior to grade against, not a leaker's guess.

GrantWhich means it's the only line item where we get to say pass or fail and mean it.

MayaSo let's see how that changes the final math.

GrantTime to add it all up. Okay, so before we close the book, Yanko Design's same rumor also had a launch date and a price for this thing. September 18th, $2,000.

Maya$2,000 for a phone?

GrantFor a folding phone, yeah. That's the number floating around.

MayaOkay, but hang on. That same piece is the one with the hinge and the cavity we already picked apart?

GrantRight. Same source, same problem. They're stating a date and a price with the same confidence they used for the engineering claim nobody's verified.

MayaSo we can't grade the price either.

GrantWe can't. It's a number, not a decision Apple's made public.

MayaAlthough two grand does track. If they're jamming liquid metal into a folding chassis, that's not a budget build.

GrantSure, it's plausible. Plausible isn't confirmed, though. That's the whole issue with this scorecard.

MayaYeah. Let's actually look at where we landed, all five calls.

GrantEvent date, green with the asterisk. Dynamic Island, ungraded. The crease, ungraded. Liquid glass, the footnote one. And now this, price and date, also ungraded.

MayaSo three out of five we literally cannot score yet.

GrantThree out of five. That's more incomplete than pass or fail.

MayaI did not expect the honest tally to look that bad.

GrantIt's not bad. It's just what grading rumors actually gets you. We weren't wrong about most of these. We just don't know yet.

MayaThere's a difference between wrong and unconfirmed.

GrantAnd that difference is the whole reason we're doing this audit instead of just reading our old notes and calling it a day.

MayaRight. Unconfirmed isn't an insult. It's just an honest label.

GrantThat's the difference the whole scorecard hinges on. We built this thing expecting a clean report card, and what we actually got was an Audit of how good our sourcing was under pressure.

MayaWhich, fine, is maybe the more useful exercise anyway.

GrantIt is, because it tells you which predictions were built on someone's track record versus which ones were just a rumor we liked.

MayaOkay, so if somebody's watching the real keynote next week with their own list...

GrantThe thing to do before you rip apart whatever Apple actually ships is ask yourself what you actually know going in versus what you're just hoping some leaker nailed.

MayaBecause it's really easy to defend a design choice using a rumor you already want it to be true.

GrantExactly. Separate the two lists first, then critique.

MayaI like that as a rule, honestly. Know your inputs before you grade the output.

GrantAnd on September ninth, we get real inputs instead of screenshots and Discord leaks.

MayaWhich means this scorecard gets rewritten anyway.

GrantIt does, but at least now we know which parts of it were ever solid to begin with. I should have checked the technicals before I went all in on the hope, huh?

MayaSo here's the takeaway. Steal this one. Before you clown on a real decision Apple ships, go back and check whether your gut call was actually informed or just a hope wearing a leaker's name tag.

GrantOuch.

MayaWe recorded this whole thing before the actual keynote, remember? Half our gourd is still hanging on people we've never met.

GrantSpeaking of calls I've never met a consequence for, remember when I swore the Ultra would launch under a grand?

MayaOh, I remember. You said it with your whole chest.

GrantI said it with my whole chest, and I was off by hundreds of dollars.

MayaEvery episode, one wrong number from Grant.

GrantIt's a tradition now, apparently.

MayaOne day I'll let you live it down.

GrantDoubtful, but I appreciate the optimism.

MayaAlright. Screenshot whatever screen made you yell at your phone this week and send it our way.

GrantSubscribe if you want the actual keynote grading next.

MayaAnd leave us a review. Tell us which prediction you think we're most wrong about.

GrantSee you on the other side of September ninth.

More episodes

Keep listening

Other episodes of UX Autopsy, newest first.

All episodes of UX Autopsy

Sources

Where this came from

7 reports behind the episode. Every one of them opens where it was published.