Two Features That Turn Speech Into Text
Apple unveiled Siri Recap and Live Rewind for the Apple Watch Series 12 and Ultra 4 on September 9, 2026, part of a broader Audio Intelligence suite alongside Sound Recognition and Shazam integration. Live Rewind lets a wearer double-press the Digital Crown to see a text version of the previous 15 seconds of conversation; Siri Recap listens through the day to generate summaries, titles, and key points, without producing a verbatim transcript or identifying who said what, according to Apple. Both features process audio in the S11 chip's Secure Exclave and, Apple states, delete the audio immediately after conversion to text; the watch sounds an audible tone, even in silent mode, and shows a visual cue to alert anyone nearby.
The features arrive in beta later in 2026, require an Apple Intelligence-enabled iPhone 16 or later, and launch in English only, according to Apple's own announcement.
The US Question: Does a Transcript Count as a Recording
Both the Register and TechCrunch centered their coverage on US wiretap and eavesdropping law, which in 11 states requires the consent of every party to a conversation before it can be recorded. Apple's defense, as the Register reported, rests on the claim that no recording exists to share, forward, or produce, since only a text summary is generated and the underlying audio is discarded. Electronic Frontier Foundation privacy director Adam Schwartz told the Register that conversational privacy must include freedom from other people using technology to document what is being said without clear opt-in consent. No US court has yet ruled on whether a text transcript generated from a live microphone falls within statutes written around audio interception.
The Question Neither US Outlet Asked
Neither the Register nor TechCrunch mentioned GDPR, the EU's wiretap-adjacent criminal statutes, or the UK's Investigatory Powers Act, and that gap matters because the underlying legal test is different in Europe. US wiretap law asks whether a recording was made. GDPR does not ask that question at all: it applies the moment personal data is processed, and a bystander's spoken words, converted into a text summary that identifies a conversation's content, are processing of that bystander's personal data under GDPR's definition, whether or not any audio file is ever saved. Apple's core defense in the US, that no recording exists, has no equivalent force under EU law, where the relevant act is the processing itself.
That leaves an open question none of the current coverage addresses: what lawful basis covers processing a third party's speech who never opened the Apple Watch settings menu, and what happens when that bystander later asks to see, correct, or delete data generated from a conversation they did not know was being converted to text. The audible tone and visual cue that satisfy Apple's stated approach to consideration for people nearby were designed against a US framing that never asks these questions.
Read next: Even Realities Hit a Billion Without a Camera | Meta's AI Built A Profile Of A Child Nobody Asked For


