Blocking every “anonymous” IP address looks like good security, but Apple’s iCloud Private Relay puts ordinary iPhone, iPad and Mac users in the same bucket as people hiding their country.
Plenty of teams on AWS switch on the managed rule group for anonymous IP addresses, leave the action on Block and move on. The reasoning is simple: a visitor behind a VPN, a proxy or a Tor node is more likely to be a bot, a fraudster or someone dodging a geographic restriction.
A growing share of “anonymous” traffic, though, comes from people who are not hiding where they are. Apple’s iCloud Private Relay, included with every paid iCloud+ subscription, sends Safari browsing through Apple’s relay network and gives websites a different IP address. To a signup form, a Private Relay user can look a lot like a VPN user. For services with location rules (streaming, banking, online gambling), telling the two apart decides whether a real customer gets in.
What Private Relay Does to a Visitor’s IP Address
Apple’s guide for engineers, Prepare your network or web server for iCloud Private Relay, explains the design. Each request passes through two relays run by different entities, so no single party, Apple included, sees both who the user is and which site the user visits. Websites see an address from Apple’s relay pool, and one address “may be shared among more than one Private Relay user in the same area.”
For anyone enforcing location rules, the key detail is that Private Relay keeps the user’s region. Apple’s guide states that “your server can trust the region assigned to the IP address it sees.” The Mac user guide lists only two location settings for Private Relay: “Maintain general location” and “Use country and time zone.” Neither setting lets the user pick another country, and choosing a country is the main reason people buy a commercial VPN.
Apple also publishes every egress address the relay uses, with its location, as a CSV file at mask-api.icloud.com/egress-ip-ranges.csv.
What the AWS Anonymous IP Rule Group Checks
The AWS documentation for the Anonymous IP list managed rule group describes two rules. AnonymousIPList covers “sources known to anonymize client information, like TOR nodes, temporary proxies, and other masking services.” HostingProviderIPList covers web hosting and cloud providers, which “are less likely to source end-user traffic.” Both rules block by default.
The AWS documentation does not mention Private Relay by name, and whether a relay address counts as an “other masking service” depends on threat data AWS does not publish. The safe assumption is that some Private Relay traffic will match, and the only way to know is to measure the traffic your own application receives.
The same documentation adds one detail: the hosting provider list “does not include AWS IP addresses.” A request from a server rented on EC2 will not trip HostingProviderIPList.
How Much Signup Traffic Gets Caught
Online gambling makes a useful test case, because licensed casinos run some of the strictest location checks on the web. The research team behind Kasinohai collected every casino registration made by visitors who clicked through to a casino from Kasinohai’s website between May 1 and July 31, 2026. The total was 3,118 signups from 2,697 unique IP addresses. Each address went through a commercial IP intelligence service of the kind operators use at signup, and the results were compared with Apple’s published relay ranges.
The commercial service flagged 137 of the 3,118 signups (4.4%) as proxy, VPN or hosting traffic. Checked against Apple’s list, 85 of those 137 flagged signups (62%) were Private Relay. The remaining 52 (38%) came from hosting providers, rented servers and commercial VPNs. And 124 of the 137 flagged signups still resolved to the player’s real country.
The same study of iCloud Private Relay in online casino signups reviewed the published terms of 174 operators in the Finnish market, and 79% of those operators explicitly prohibit VPN use. Some clauses name any “anonymisation service” or “location-masking software,” wording broad enough to cover Private Relay. The study found no public case of an operator citing Private Relay as the reason for freezing an account. For engineers, the finding is concrete: one “anonymous” flag merged two very different groups, and the larger group was Apple device owners using a privacy feature they pay for.
A Better Setup on AWS
Separating the two groups takes a handful of changes.
1. Override the AnonymousIPList rule action to Count. The rule group still adds the label awswaf:managed:aws:anonymous-ip-list:AnonymousIPList to matching requests, so later rules in the web ACL, or the application itself, can decide what happens next.
2. Keep HostingProviderIPList on Block for signup and login paths. Real customers rarely sign up from a data center, and the 52 hosting and VPN signups in the Kasinohai data are the traffic the rule was written for.
3. Load Apple’s egress list on a schedule. An Amazon EventBridge schedule can trigger a Lambda function daily to fetch the CSV, keep a dated copy in S3 and refresh a lookup table. Check the size of Apple’s CSV before planning to load the ranges into a WAF IP set, because AWS WAF allows 10,000 addresses in CIDR notation per IP set and the quota cannot be raised. A longer list belongs in the application layer, for example a DynamoDB table keyed by network prefix.
4. Rate limit relay addresses as shared addresses. Apple’s guide recommends treating relay IPs like carrier-grade NAT. A per-IP limit tuned for home broadband will lock out unrelated customers who share a relay.
5. Swap some CAPTCHAs for Private Access Tokens. Recent Apple devices support Private Access Tokens, which let a server confirm a request comes from a genuine device and account without a puzzle.
Keep the Fraud Logs Lawful
Under GDPR an IP address can count as personal data. A table of Apple’s relay ranges is not personal data, but a log joining relay addresses to customer accounts needs a stated purpose and a retention period. An earlier myAWS article on the challenges of maintaining a GDPR-compliant cloud platform covers retention and purpose limitation in more detail.
Final Thoughts
Before tightening any rule on anonymous traffic, split the flagged requests by Apple’s published ranges and count how many genuine customers sit in the relay group. In the Kasinohai signup data, Private Relay accounted for 85 of the 137 flagged signups. Count mode in AWS WAF, Apple’s CSV and a daily Lambda function are enough to see the same split in your own traffic.
