WWDC 2026: Everything Apple Announced (Siri Now Runs on Gemini) artwork

WWDC 2026: Everything Apple Announced (Siri Now Runs on Gemini)

Elon Musk Podcast

June 9, 2026

Apple rebuilt Siri from the ground up at WWDC 2026, and the surprise is what's under the hood: Google Gemini. The newly renamed "Siri AI" gets a standalone app and real-time screen awareness, with Gemini-based models integrated into Apple Intelligence through Private Cloud Compute.
**SPEAKER_1** (0:00)
Apple's overhaul of Siri is actually being powered behind the scenes by Google's Gemini models. Tim Cook announced the partnership during his final keynote before handing the CEO position over to John Ternus.

**SPEAKER_2** (0:11)
Yeah, you are basically looking at the ultimate closed loop company, quietly opening a backdoor to their biggest rival.
Apple spent well over a decade building an ecosystem defined by strict internal control over both the hardware and the software. So relying on Google's foundation models to power the most personal digital assistant on their devices just alters that approach completely.

**SPEAKER_1** (0:36)
It really does. And it makes you wonder if integrating a competitor's artificial intelligence engine into the core of their operating systems compromises the privacy and the hardware control they are known for, or if it was really just the only way they could catch up.

**SPEAKER_2** (0:49)
I think that tension pretty much defines this entire software generation. They are attempting to bolt a highly advanced and very data-hungry language model onto an operating system that was explicitly marketed on the premise of keeping your data locked down.

**SPEAKER_1** (1:02)
Yeah, exactly.

**SPEAKER_2** (1:03)
The engineering required to make those two philosophies co-exist is incredibly complicated.

**SPEAKER_1** (1:09)
We should probably look at how this integration actually manifests on the devices themselves. Because Siri has been completely rebuilt from the ground up. They are officially calling it Siri AI now.

**SPEAKER_2** (1:20)
Right. The whole visual presentation is entirely different now.

**SPEAKER_1** (1:22)
It is, yeah. On devices with a dynamic island, the old glowing orb at the bottom of the screen is just gone. Instead, there is this semi-transparent glowing waveform that lives inside the pill-shaped cutout at the top of the display.

**SPEAKER_2** (1:38)
Moving that visual feedback to the top of the screen really changes the ergonomics of the interaction. Because the old Siri hijacked the entire bottom half of your display, which constantly interrupted whatever you were doing.

**SPEAKER_1** (1:51)
It pulled you completely out of your context.

**SPEAKER_2** (1:53)
Exactly. Confining the animation to the dynamic island allows the assistant to feel a bit more like a background utility rather than a foreground application.
It just sits up there glowing and pulsing while you continue to scroll a web page or read an email.

**SPEAKER_1** (2:06)
And on the iPhone, you can also swipe down from the middle of the screen to pull up a new card-based interface for text input. But I think the most significant functional change here is the introduction of a standalone Siri application.

**SPEAKER_2** (2:20)
The dedicated app.

**SPEAKER_1** (2:21)
Right. This app uses iCloud to privately sync your entire conversational history across your phone and your computer and your headset.

**SPEAKER_2** (2:29)
A persistent memory bank fundamentally changes the nature of a voice assistant.
Previously, Siri was a single-command utility. You asked it to set a timer for pasta or you asked for the weather, and the interaction ended the exact moment the task was completed.

**SPEAKER_1** (2:44)
The software had no memory of what you asked of five minutes prior.

**SPEAKER_2** (2:47)
Zero memory. By moving Siri into a dedicated application with a synced history, it transitions into a continuous chatbot experience.

**SPEAKER_1** (2:55)
So you can start a research query on your Mac and then leave your desk and pick up the exact same conversational thread on your Apple Watch or Vision Pro while you are walking down the street.

**SPEAKER_2** (3:04)
Yeah.

**SPEAKER_1** (3:04)
It treats your interactions as an ongoing relationship rather than a series of isolated commands.

**SPEAKER_2** (3:10)
I really have to question whether users actually want a saved syncable log of every mundane question they ask their devices though.

**SPEAKER_1** (3:19)
You think it's too much.

**SPEAKER_2** (3:20)
I just think we have been conditioned to view voice queries as ephemeral. You ask the room a question and the machine answers and the audio just dissipates.
Creating a written transcript of those interactions introduces a subtle friction.

**SPEAKER_1** (3:34)
There is definitely a psychological difference between talking to a temporary assistant and contributing to a permanent digital record. When you ask a digital assistant how many ounces are in a cup while you are cooking, you do not necessarily need that query archived and time stamped and synced to your laptop.

**SPEAKER_2** (3:52)
Exactly. Users might start self-censoring or feeling self-conscious about the simplicity of their questions if they know those questions are being saved to a permanent log.

**SPEAKER_1** (3:59)
Right.

**SPEAKER_2** (4:00)
It forces the user to confront their own digital footprint in a way that previous iterations of Siri sort of avoided. You are essentially looking at a search history of your own immediate unfiltered thoughts.

**SPEAKER_1** (4:10)

18 more minutes of transcript below

Feed this to your agent

Try it now — copy, paste, done:

curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Works with Claude, ChatGPT, Cursor, and any agent that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire.

Using your own key:

curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/1000771890783