Apple adds an app-exclusion setting for Siri AI

Apple adds an app-exclusion setting for Siri AI

The Electronic Frontier Foundation says Apple added an “Exclude Apps” setting for Siri AI in iOS 27.2, following a request from the EFF campaign. The setting lets users prevent Siri AI from accessing selected apps, but some features stop working when an app is excluded. EFF says this trade-off can help users protect app contents, including those in secure messaging apps, from AI access. It says the feature failures show the setting is effective and welcomes the change. EFF contrasts Apple’s per-app control with Google’s Gemini settings, which it says are less straightforward and do not offer per-app control of screen context. EFF says Gemini may still access content through notifications or screen context, depending on settings. It calls for Google to offer per-app controls. EFF also urges operating systems to consider similar controls and says Apple should offer an option to use “on device” AI features only. The article describes the controls and their possible privacy implications; it does not report independent testing of what data each system accesses.

Publication: 6289c269-6a27-40a9-85bb-f82d19da1855

attribution — pass: The reading consistently attributes the account, assessment and asks to the Electronic Frontier Foundation (EFF), the publisher. The Apple and Google settings are presented as EFF's description, not as established fact or Escape Meridian's opinion.

fulfilling one of our asks from our Encrypt It Already campaign earlier this year

summary — uncertain: The draft drops EFF's central stake: many AI features need cloud processing, so secure-messaging content may leave the device without the user knowing. Without this, 'protect app contents from AI access' and the on-device ask are hard for a general reader to follow. Screen context and on-device AI are also unexplained. Smallest repair: add one sentence on the cloud-processing concern and add tooltips for those terms. The draft also loses EFF's emphatic 'That's a good thing, actually' framing of the trade-off; restoring it would keep the author's force.

Many AI features require cloud processing, and despite privacy protections engineered into these systems, the content of secure messaging apps may leave a user’s device without their knowledge.

date — pass: originalPublished (2026-10-28T13:04:21-07:00, from the publisher page) matches the feed timestamp of 2026-10-28T20:04:21Z, and the article refers to the current iOS 27.2 update. No earlier first-publication date is indicated.

support — uncertain: Claim 4 generalizes too far. EFF directs the per-app ask at Google and the on-device-only ask at Apple, and says only that all operating systems 'need to consider similar controls'. Smallest repair: attribute each ask to its target. The limitation that EFF reports no independent testing is also arguable, since EFF describes its own hands-on checks (Signal exclusion messages; Gemini 'seems to work well enough'). Repair: say EFF reports its own use of the settings but not independent verification of data flows.

But we think they should go further and offer the option to only use “on device” AI features, similar to what Samsung offered with its Galaxy AI.

media — pass: The selected media is the verified direct link to the original EFF article. This is the most relevant way to reach the source; the alternative banner GIF is decorative.