How we rate these claims
Three ratings, applied the same way every time. Strong means either multiple independent reports converge on the same detail, or the claim is a direct continuation of a path Apple is already visibly on — Liquid Glass maturing in its third year, for instance, isn't a leak, it's a pattern. Rumor means the opposite: a single report, thin or early sourcing, or a claim that's plausible mainly because it fits Apple's general direction rather than because anyone has confirmed specifics.
Confirmed is the strict one. It is reserved exclusively for Apple's own announcements — a keynote, a press release, a developer document — and nothing on this page can earn it before the WWDC 2027 keynote on June 7, 2027, no matter how strong the reporting gets beforehand. We'd rather leave a well-sourced claim at Strong for a year than mislabel it.
One more thing worth stating plainly: we do not name individual leaker accounts or outlets we can't independently verify. Claims here are framed as "reporting suggests" or "Apple's historical pattern shows" rather than attributed to a specific handle, because we'd rather be vague and honest than specific and unaccountable. See our about and methodology page for how this whole site is built and updated.
The next-generation Siri Strong
The clearest storyline heading into iOS 28. Apple has spent three years rebuilding Siri around large language models, shipping the overhaul in stages — and reporting consistently points to iOS 28 as the version where the assistant becomes genuinely conversational and agentic: able to see what's on your screen, act across apps ("find the ferry photos from Ali, fix the exposure, send them to Mom"), and hold context across a whole conversation.
The open questions are hardware and rollout. On-device models of this size want RAM, so expect the full experience to be gated to recent iPhones — see Apple Intelligence requirements — and expect the most ambitious pieces to arrive in iOS 28.1 or later, as Apple has repeatedly staged its AI launches rather than shipping everything on day one.
The reasoning behind the Strong rating is infrastructure, not speculation. App Intents, the framework that lets apps expose their actions to Siri and Shortcuts, has been Apple's developer-facing priority for two straight cycles, and it's the plumbing an agentic assistant needs to actually do things rather than just answer questions. iOS 28 is the version that investment is supposed to pay off in: an assistant that chains several app actions together instead of handling one request at a time.
For that to work at the level Apple has described, App Intents adoption needs to be broad enough that Siri has something to act on in the apps you actually use day to day, and the on-device models handling the sensitive parts of a request need enough Neural Engine headroom to run without noticeable lag. That's the practical reason the deepest features skew toward recent hardware, and it's also why Apple has a track record of splitting ambitious AI features across point releases — as it did with parts of Apple Intelligence across iOS 18.1, 18.2, and 18.4. iOS 28.0 and iOS 28.1 could look meaningfully different for Siri.
Liquid Glass 3.0 Strong
iOS 26 introduced Liquid Glass, the biggest visual redesign since iOS 7, and iOS 27's cycle has been about settling it down. Third iterations are historically where Apple's design languages mature — expect iOS 28 to push the material deeper into the system: more adaptive translucency, richer depth on the Lock Screen, and widgets and Control Center surfaces that respond to content behind them. Less a new look, more the current look finishing the job.
Apple's design languages typically take about three cycles to settle. The flat design introduced in iOS 7 was rough at launch and kept getting refined through iOS 9 and iOS 10 before it felt finished; Liquid Glass looks to be following the same arc — introduced in iOS 26, adjusted for readability and contrast feedback through iOS 27, with iOS 28 positioned as the release where the material stops visibly changing shape and starts getting used more confidently across the system.
Because this is a software change rather than a hardware-gated one, it should reach every device that runs iOS 28 at all — see the full supported devices list — not just the recent iPhones the AI features on this page require. The real risk isn't that it fails to happen, it's that the changes are incremental enough that most people notice a system that simply feels more finished than iOS 27, rather than a single new headline feature.
Live translation everywhere Rumor
Apple has been assembling translation pieces for years — on-device Translate, live call translation, camera text translation. The iOS 28 rumor is unification: one system-wide translation layer covering calls, Messages, FaceTime, and anything the camera sees, all processed on-device, with an API so third-party apps get it free. It fits Apple's privacy-first pattern perfectly, which is exactly why it should still be treated as a rumor: it's as much inference from trajectory as leak.
The pieces already exist and have for a while: the standalone Translate app since iOS 14, live translation during Phone and FaceTime calls, and Camera-based text translation. Apple rarely maintains this many overlapping single-purpose tools without eventually merging them into one system-level layer — the same pattern that turned separate dictation and writing features into the single Apple Intelligence pitch.
Making translation system-wide and instant needs on-device models fast enough for live conversation, which likely means it inherits the same Apple Intelligence device floor as Siri rather than reaching every iPhone that gets iOS 28. It's also plausible Apple ships this in stages — Messages and calls first, camera and a third-party API later — rather than all at once, which is exactly why it stays a Rumor rather than Strong for now.
AI privacy dashboard Rumor
As AI features multiply, so does the question of what they're reading. The rumored answer: a single dashboard — think App Privacy Report, but for intelligence — showing which AI features accessed which data, what ran on-device versus on Private Cloud Compute, with per-feature kill switches. Thin sourcing so far, but it's the kind of trust-building feature Apple loves to keynote against its competitors.
Apple already has the scaffolding for this. App Privacy Report has shown which apps access your data since iOS 15.2, and the Apple Intelligence rollout added its own disclosures about what runs on-device versus what's handed to Private Cloud Compute. Folding those into one dashboard is a small conceptual step, not a new capability Apple would need to invent from scratch.
What's missing is specific sourcing — this reads more like an educated guess at how Apple would package something it already has the pieces for than a leak about a feature actually in development. If it does ship, expect it inside Settings > Privacy & Security rather than as a standalone app, and expect Apple to frame it explicitly around Private Cloud Compute's on-device-versus-cloud distinction.
A generative keyboard Rumor
Beyond autocorrect: context-aware sentence completion in your own writing style, inline tone rewriting without detouring through Writing Tools menus, and smarter multilingual switching — all on-device. Reports here are early and vague; we're keeping it at Rumor until the sourcing improves.
Apple already ships two components this would combine: Writing Tools, which rewrites and proofreads text on-device since iOS 18, and predictive text that already adapts to how you personally type. A keyboard that applies Writing Tools-level intelligence inline, without a detour through a menu, reads as a logical next step rather than a new direction for Apple.
The real constraint is memory. Third-party keyboard extensions run in a tightly sandboxed process with a strict memory ceiling, specifically so they can't slow down or destabilize the rest of the system — and a genuinely generative keyboard wants more headroom than that sandbox typically allows. That's a concrete reason this could stay limited to Apple's own first-party keyboard for a while, or arrive later than the rest of the Apple Intelligence features on this page.
Local & NAS backups return Rumor
A perennial power-user wish with fresh smoke around it: encrypted iPhone backups directly to an external drive or network storage, no Mac or iCloud subscription required. It would sell fewer iCloud plans, which is the argument against; the argument for is the EU's growing pressure on ecosystem lock-in. Genuinely could go either way.
The case for it has gotten stronger over the past two years, not weaker: the EU's Digital Markets Act has already forced Apple to allow alternative app marketplaces and browser engines in Europe, and local network or external-drive backup is a smaller ask by comparison — Android has offered it for years without much controversy.
The case against is just as concrete: iCloud+ subscriptions are recurring revenue, and a free local backup option competes directly with them. If this ships at all, the same regulatory pressure that forced sideloading into the EU only makes a Europe-first or Europe-only rollout more likely than a global one. Worth tracking, but don't expect it everywhere even if it happens somewhere.
Smaller changes in the pipeline
- Battery intelligence — AI-driven charge scheduling that learns your week, not just your night. Rumor
- Messages — continued RCS upgrades and smarter thread summaries. Strong
- Health — the AI health coach reportedly in development is a candidate for an iOS 28-era debut, possibly tied to new hardware. Rumor
- CarPlay — deeper instrument-cluster integration continuing its slow rollout. Strong
- Default apps — more user-selectable defaults, driven by EU regulation. Strong
What iOS 27 tells us about iOS 28
As of today, iOS 27 is in beta and its final release is due around September 2026 — which makes it the best evidence available for what iOS 28 will look like, more useful than any single leak. Apple doesn't design each release in isolation; every cycle seeds the next one in three predictable ways.
First, features that ship unfinished get completed the following year. That's exactly what happened with Apple Intelligence itself, which arrived incomplete in iOS 18 and kept gaining pieces through iOS 18.4 and into iOS 26 — and it's the same lens to use on whatever iOS 27 ships half-built. Second, features Apple announces but delays roll forward rather than disappearing; a capability previewed at WWDC 2026 that slips past iOS 27's launch is a strong candidate to headline iOS 28 instead. Third, design languages iterate rather than reset — Apple has never followed a major visual overhaul with another major visual overhaul the very next year, which is a big part of why Liquid Glass 3.0 reads as refinement rather than replacement.
Put together, iOS 28 should look less like a clean slate and more like iOS 27 with the rough edges sanded off and the delayed pieces filled in — consistent with how Apple has run every release cycle in recent memory. For the longer history behind why Apple's version numbers jump the way they do, see why Apple skipped iOS 19 through 25.
What we're hoping for (not predicting)
Filed honestly under wishes, because no reporting supports them yet: real Lock Screen widget interactivity, per-app notification AI summaries that can be tuned rather than toggled, Shortcuts rebuilt around the new Siri instead of alongside it, and — the eternal one — a clipboard manager.
Features that probably won't happen
Every wishlist has entries that keep not arriving, year after year, for reasons that have nothing to do with technical difficulty. Our full iOS 28 wishlist keeps wishes separate from predictions; this section is the other side of that coin — the structural reasons a few perennial requests are unlikely to show up.
- True file-system access and desktop-class multitasking outside the EU — the Files app and multitasking keep improving, but unrestricted file access cuts against the sandboxing model iOS security is built on. Where Apple has loosened this at all, it's been confined to the EU under regulatory pressure, not offered globally.
- Sideloading apps outside the EU — already live in Europe because the Digital Markets Act required it, with no sign Apple intends to extend it elsewhere voluntarily. App Store commissions are too large a revenue line to give up without being forced to.
- Fully customizable Home Screen icons — iOS 18 added tinting and light/dark icon themes, which was Apple's answer to years of demand, but true third-party icon replacement would mean reopening APIs that have stayed closed for a decade for stability and consistency reasons.
- A system ad blocker that also blocks Apple's own ads — Apple runs a real, growing ad business inside the App Store and Apple News. A privacy feature that happened to cut into that revenue is a much harder sell internally than one that only affects competitors.
When we'll know more
Rumor season follows a rhythm: scattered supply-chain and analyst reports through the autumn, the first substantive leaks in spring 2027, a credible feature picture by May, and the truth on June 7, 2027. This page updates at every step — each claim above carries its rating's date, and anything Apple confirms moves to Confirmed within the hour. And when it all becomes real, the beta guide is how you try all of this yourself next June.