With the Apple Watch Series 12 and Ultra 4, Apple is introducing some neat new audio-based AI features that seem genuinely useful and exciting.

The new S11 processor includes hardware that makes it possible to continually listen and (lightly) process audio without ruining battery life, and Apple is using that to provide four new AI-powered features. It calls these features “Audio Intelligence” because Apple can’t help but brand everything.

• Full-time Shazam: The watch can detect music playing and automatically identify it using Shazam, displaying it on widget in the Smart Stack. Apple Watches already do this on demand, but the new thing here is that it does it continuously in the background like Pixel phones.
• Sound recognition: The same feature you can get on HomePods and iPhone now comes to Apple Watch. You can get notified visually when your watch hears sounds such as doorbells, fire alarms, sirens, etc. You used to need your iPhone nearby to use this feature, but the Series 12 and Ultra 4 can do it independently.
• Live Rewind: Double-tap the digital crown to see a text readout of the last 15 seconds of what was said around you. You can then ask Siri about it or save it to the Siri app.
• Siri Recap: Get a high-level summary of an entire conversation in the Siri app, so you don’t forget specifics or need to take notes.

The first two of those features don’t really bother anyone. Those are existing features, and the only new thing is that the Apple Watch can do them in the background without iPhone intervention. The last two will be available later this year as a beta, and those are the ones that have people freaking out. Are people’s Apple Watches going to be recording you all the time?

Apple has issued a Audio Intelligence Security Overview white paper to help users understand how all this works and how it protects everyone’s privacy. It’s worth a read, but here are the highlights.

How Live Rewind works

On the Apple Watch Series 12 and Ultra 4, audio from the microphone is continuously flowing into a little buffer in the Secure Enclave. This actually happens on lots of Apple products; it’s how they know if you’ve said “Hey Siri” or not. This is a rolling buffer, where new data continuously replaces old data.

The Secure Enclave is a completely separate hardware block that is not accessible to the rest of the operating system, or any apps, or even Apple. It’s where your Face ID and Touch ID data is stored, for example.

Apple

When you double-press the Digital Crown, the last 15 seconds of that buffer is sent via an encrypted secure transfer to your iPhone, where an AI model transcribes it. This happens in the iPhone’s Secure Enclave, so the audio cannot be accessed by the rest of iOS, or any app, or the user, or even Apple. This is just a straight voice-to-text transcription; it doesn’t identify or label people who are speaking. The audio is then immediately discarded.

The text of the translation is then sent back to Apple Watch where it appears on the display. It is not saved unless you ask Siri about it or save it to the Siri app to revisit later. The Siri app’s data is end-to-end encrypted and synced between devices, and users can control whether conversations are automatically deleted after 30 days, 1 year, or saved forever. Apple cannot decrypt this data, even with a court order—only the user has the key.

And one more thing, an audible chime plays from the speaker on your Apple Watch when you implement Live Rewind, even if your Apple Watch is on silent or you have headphones connected.

So that’s how Live Rewind protects your privacy and those around you. Raw audio data can’t ever be accessed by the user, the system, Apple, or any apps. It’s just straight translated to text, which is then optionally saved in the end-to-end encrypted Siri app.

How Siri Recap works

Siri Recap works in a similar way, but there’s more going on. You’ll be able to leave it on all the time, only when you trigger it manually, or specify certain places or times when it can’t operate.

First, a tiny little lightweight model simply detects if a conversation is happening at all.

If it is, that encrypted and secure audio buffer is sent to the iPhone (via an encrypted data transfer) for processing. It is transcribed to condensed text, and the raw audio is then immediately discarded. It is never accessible to the OS, apps, the user, or Apple.

This transcribed text passes through a safety filter than will remove potentially harmful terms. Then it is encrypted and sent to Apple’s Private Cloud Compute servers to be summarized, along with relevent contextual data that can make the summary better (such as what music you might be listening to, a relevant calendar entry, or high-level place information like what city you’re in).

Apple

Private Cloud Compute generates a title and summary with key points. It removes any specific identifiable or sensitive info such as place names or financial data, while leaving generic places names like “park” or “grocery store.” That summary is then sent back to your devices, encrypted, where it is stored in the Siri app in a new Recaps tab. All data on Apple’s Private Cloud Compute servers is always discarded immediately after processing; Apple never saves any of it.

The summaries in the Recaps tab are deleted automatically after seven days unless you choose to save it.

Like Live Rewind, Siri Recap does not create a recording, does not identify speakers, and does not even create a verbatim transcript. No audio is ever accessible at any step of the process by the user, any app, or Apple.

Finally, Live Rewind and Siri Recap are not available for users under the age of 13. For users under 18 on a Child Account, access to Live Rewind and Siri Recap must be turned on by a parent or guardian.

OpenAI has paused new sign-ups and upgrades to its $200 ChatGPT Pro tier, citing a surge in demand for its Astra capability that is placing pressure on system capacity, according to company statements and an executive post on X.

“To make sure our current users have an incredible experience and continued access to Astra, we are going to pause subscriptions to our $200 Pro plan,” OpenAI member of technical staff Thibault Sottiaux wrote in a post on X. “These put the most strain on our systems and we wanted to take the smallest step that allows us to continue giving the broadest access possible.”

The company separately confirmed the move in its help documentation, stating that “as of September 10, 2026, we’re temporarily pausing new sign-ups and upgrades to the ChatGPT Pro $200 plan (Pro 20X).” The pause applies to users across Free, Go, Plus, and Pro $100 tiers seeking to upgrade, while “existing ChatGPT Pro $200 subscriptions … are not affected by this pause,” it said.

Sottiaux added that “there is no impact to existing accounts and we are working on adding more capacity as fast as we can,” pointing to ongoing efforts to scale infrastructure in response to demand.

Users who cancel or downgrade during the pause will not be able to re-subscribe to the $200 tier until the restriction is lifted, according to the company’s help page. The $100 Pro tier remains available, with lower usage limits. Promotions tied to the $200 tier are also paused, the page read.

Astra demand drives capacity decisions

The decision is directly tied to the uptake of Astra, which OpenAI has positioned as a more advanced capability within its platform. The scale of that demand, according to Sottiaux, has been atypical.

“Demand for Astra is really unprecedented,” he wrote in the post. “We’re pulling all the levers possible to sustain the demand, but I’ve not seen anything like it until now and we went through very steep growth before.”

He added that “priority will always be to keep excellent service for existing users,” noting that the company “might have to pause new Pro subscriptions for a bit if this continues.”

The pause comes amid broader rollout challenges tied to Astra. OpenAI Chief Executive Sam Altman recently described the launch as “messy” after some paying users were unable to access the model immediately, reflecting the operational complexity of deploying high-demand AI systems at scale.

Capacity constraints surface for enterprise workloads

The pause, analysts said, reflects how demand for high-intensity AI workloads is intersecting with infrastructure limits, particularly at the highest usage tiers.

“It signals that frontier capacity is still rationed,” said Bhupendra Chopra, chief revenue officer at Kanerika. “One week after launching Astra, OpenAI paused new sign-ups and upgrades to the $200 Pro tier, the plan with the heaviest Astra usage limits, while the $100 Pro tier, the API, Business and Enterprise all stayed open. That tells you the priority order.”

“Consumer power users are the release valve. Enterprise contracts are what the vendor protects,” Chopra said. “For CIOs the lesson is that a model being announced and a model being available to your workloads at the volume you need are two different events.”

Chopra said the move reflects how access to advanced AI capabilities is being managed.

“We are already there,” he said. “Rate limits, usage caps, queueing and now sign-up pauses are all forms of capacity gating, and every frontier vendor uses some of them.”

“A subscription buys you a place in the line. A fixed slice of compute comes only from a contract that says so,” he added.

Planning for constrained supply

For enterprise IT leaders, Chopra said model capacity should be treated as a dependency.

“Treat model capacity like any other supply chain dependency,” he said. “Ask the vendor three questions before you sign. What throughput is contractually committed, what happens to my workloads when demand spikes, and how quickly can I fail over to another model.”

“If a workload is production-critical, it should run on an enterprise agreement with committed capacity… and it should have a second model tested and ready,” he said.

A recurring pattern

Chopra said the pause reflects a broader pattern seen in recent launches.

“This is at least the third time in two years OpenAI has throttled or paused access at launch,” he said. “Each launch outruns capacity, capacity catches up, and the next model outruns it again.”

“OpenAI calls this pause temporary, and it will lift once capacity catches up,” he added. “The structural condition… will persist as long as model capability keeps outpacing data center buildout.” OpenAI has not provided a timeline for when new sign-ups to the $200 Pro tier will resume, stating only that the restriction is temporary.

The article originally appeared on CIO.com.