Apple will monitor your public iCloud links now, starting with iOS 27.3

Apple plans to begin monitoring iCloud links that users share publicly, with the change arriving alongside the 27.3 build of its operating systems across its platforms. The company says it will scan such links for harmful or potentially illegal content and reserves the right to act against accounts that break its policy. The policy covers only shares set to "Anyone with the link," the designation used when someone posts an iCloud link in a group chat, a Discord server or anywhere else strangers can reach it. Apple's announcement did not define what counts as harmful content in practice.
The enforcement menu Apple outlined is broad. It says it may move, refuse, modify, remove or delete content at any time. It may block or disable a user's account, and it may block the ability to create a new account tied to the same email address or phone number. Apple also lists involving law enforcement as a possible step.
The encryption detail is the part developers and privacy-minded users should focus on. When an iCloud link is shared under the "Anyone with a link" setting, end-to-end encryption is removed, even if either party has turned on Advanced Data Protection, which is Apple's optional setting for extending end-to-end encryption to more iCloud data. That removal is what lets Apple view the link's contents at all. Sharing with "People you choose," which restricts access to invited or approved users, keeps encryption in place, and Apple says it cannot see what is shared that way. In other words, the technical distinction between the two sharing modes is the whole ballgame: one is readable by Apple, the other is not.
That framing helps explain how the system can work without Apple reading private shares. End-to-end encryption, in the ordinary sense, means the service provider holds keys it cannot use to decrypt content; removing it for public links converts those shares into something closer to content stored on a server the company can inspect. A likely reading is that scanning happens server-side on links that are already unencrypted by design, rather than on-device or against private shares. Apple has not described the detection method, so whether this is automated matching against known material, human review triggered by reports, or some mix is not established by the announcement.
The definition gap is the practical concern. Obvious categories such as drug sales and child sexual abuse material would presumably fall inside any such policy, but the phrase "harmful content" as used here is left open. For a developer or creator who routinely posts public iCloud links, that ambiguity translates into uncertainty about where the line sits. The policy also does not lay out an appeals process for people who believe their account or content was wrongly flagged. Apple does maintain a web page describing how to appeal account actions for other reasons, so a framework for pushing back exists even though this policy does not reference it directly.
This is not the first time Apple has waded into content scanning, and the history is worth keeping in mind as context. The company's earlier proposal to scan photos on devices for known child abuse imagery drew sustained criticism from security researchers and privacy advocates, who argued that client-side scanning could be repurposed and would weaken trust in on-device processing; Apple shelved that approach. The current change is narrower in one respect, because it targets shares the user has deliberately made public and unencrypted, but it lands in a similar debate about how much inspection users should expect from a storage provider.
For product teams, the immediate questions are operational. Anyone building features that hand out iCloud links, such as shared media galleries, support attachments or collaborative documents, should know which sharing mode the flow selects, because that choice now determines whether Apple can see the content and whether the account behind it carries enforcement risk. Tools that generate "anyone with the link" URLs to reduce friction for recipients are exactly the ones exposed. Defaulting to invited-access links avoids the scan but adds a step for end users; that is a real trade-off, not a free one.
Why it matters: Apple account holders who post public iCloud links now face account-level consequences, including possible bans and blocks on re-registering with the same email or phone number, while private shares remain encrypted and unscanned. Developers whose products generate or distribute iCloud links should audit which sharing mode they use by default, since that single setting decides whether content is visible to Apple. The absence of a stated appeals path for this policy is a risk worth flagging to users, and the undefined scope of "harmful content" is likely, in my reading, to invite pressure for clearer rules before the 27.3 rollout reaches everyone.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
Under Apple’s new policy, the company will scan publicly shared iCloud links for harmful content – and possibly ban violators.