Apple memory holed its broken promise for an OCSP opt-out

Apple’s online certificate status checks in macOS allow the company to see which signed apps are run from a given IP, and a previously promised system preference to disable this logging appears to have been quietly removed. Commenters weigh the security benefits of real-time certificate revocation against the privacy cost of sending app usage data to Apple, comparing it to similar trade-offs in Firefox’s OCSP handling and DNS. Many argue that relying on host firewalls, Pi-hole, or switching to Linux is now the only reliable way to avoid this tracking, while others note that any proprietary OS vendor ultimately controls user privacy through updates.

Blocking OCSP and Network Workarounds

  • Users discuss blocking Apple’s OCSP servers (ocsp2.apple.com, ocsp3, etc.) via Little Snitch; regex-style patterns are suggested if supported.
  • trustd is identified as the macOS process making OCSP requests.
  • Some note Apple apps and certain networking setups (e.g., bridged VM adapters) can bypass Little Snitch.
  • /etc/hosts is reported as unreliable for blocking, especially with Safari and iCloud Private Relay; behavior around IPv4/IPv6 is debated and partly unclear.
  • Turning off all network radios prevents OCSP checks but is seen as impractical.

Little Snitch, DNS, and Alternatives

  • Criticism that Little Snitch performs DNS resolution before the allow/deny dialog, making it weak against DNS-leak concerns.
  • A Pi-hole-based, DNS-layer blocking setup is proposed as a more robust alternative for network-wide control.

What OCSP Reveals and Risk to Typical Users

  • OCSP lets Apple see which signed apps (by developer certificate) are run from a given IP, historically in plaintext, now encrypted but still visible to Apple.
  • It’s claimed the request fires on each app launch, not just install, implying a potential (IP, app, timestamp) log.
  • One commenter questions how worrying this is for a typical user who mostly uses App Store and Homebrew; others suggest it’s more of a privacy-model concern than an immediate practical risk.

Apple’s Privacy Posture and Broken Promises

  • Strong criticism that Apple markets privacy but treats it as “privacy is when you trust Apple,” and that walking back the OCSP opt-out promise undermines trust.
  • Counterpoint: among large tech firms, Apple is seen by some as comparatively more privacy-oriented and investing in privacy tech, even if imperfect.
  • Others argue that “only one company has all your data” should not be called privacy.

Homomorphic Encryption Discussion

  • Apple’s use of homomorphic encryption for iOS 18 live caller ID lookups is noted; an SDK was announced.
  • Some see this as promising privacy tech Apple is pushing; others say it’s narrow, early, and can’t excuse not following through on simpler privacy commitments like OCSP controls.
  • It’s highlighted that homomorphic encryption is required for third‑party caller ID providers; Apple’s own features still largely rely on trusting Apple.

Security Architecture: OCSP vs Notarization

  • Clarification that OCSP checks revocation of Developer ID certificates via trustd, while notarization is a separate process using tickets signed by Apple and checked by syspolicyd/Gatekeeper.
  • Some initially conflate the two; others emphasize they are distinct security layers, both contributing to macOS’s code-trust model.

Comparisons with Other Ecosystems

  • Some users report abandoning Apple entirely over issues like OCSP and preferring all‑Linux environments, where tools like SSH don’t involve vendor-level tracking and “just work.”
  • Others note macOS also has built-in SSH but warn that enabling services without strong auth (e.g., keys/certs) can be risky.