Apple's $250M Siri settlement is a preview of the next class action shape.
A supplementary $155 million payment was added to the Lopez v. Apple class settlement in May 2026, bringing the total consideration to roughly $250 million. The suit alleged that Siri activated without a wake phrase and recorded portions of conversations, some of which were then reviewed by contractors as part of Apple's quality grading program. Apple did not concede on liability and reaffirmed that Siri data was not used to build advertising profiles.
The interesting part is procedural. Plaintiffs' counsel argued that Apple could not produce a per interaction log showing what triggered any given activation, which policy applied, or which downstream system saw the recording. The court's willingness to certify on that basis is what most defense counsel are quietly rereading. It is the template for the next round of AI privacy actions.
Bloomberg coverage →Privacy defenses are moving from policy documents to per interaction records. A company statement that data was not used a certain way is worth less than the log that shows what actually happened to each recording, each embedding, each downstream call.
Every enterprise deploying a voice or vision AI capability is now on notice. The question in discovery will not be what your policy said. It will be what your record shows. If the record is a screenshot and a vendor promise, the settlement number is a matter of arithmetic.
- capture.event: trigger phrase, ambient context, activation policy
- data.classification: PII, biometric, location, retention window
- downstream.systems: every service that saw the payload, with allow / transform / block
- consent.state: the exact consent version active at capture
- export: audit ready for regulator, plaintiff, and D&O carrier on one query
Have a voice, vision, or agent product in field?
Thirty minutes on the interaction level record you would need in discovery.
Book a working session →