Overall impressions of the guide and platform security
- Many find the 262‑page guide “hardcore” and technically impressive, especially the evolution of SoC-level security and use of features like MIE/MTE.
- Some ask how fully iOS/macOS already leverage new hardware protections; others report that major allocators and system processes on recent OS versions are already MIE-enabled.
Pegasus, Lockdown Mode, and Apple’s threat model
- A major criticism: the guide doesn’t explicitly address Pegasus-class mercenary spyware, seen as “the elephant in the room.”
- One camp argues this is fine: Pegasus just exploits normal bugs, and the document’s mitigations apply; Apple discusses these attacks in external talks and created Lockdown Mode to harden high-risk users.
- The opposing camp says Apple markets iOS as eliminating whole risk classes via its locked-down model; failing to square that narrative with real-world Pegasus attacks is “security by obscurity.”
- Debate over Lockdown Mode:
- Supporters: it significantly reduces attack surface (e.g., iMessage URL handling) and is documented in the guide.
- Skeptics: if Pegasus can bypass the main model, there’s no clear reason it couldn’t bypass Lockdown; without verifiable evidence of classes of exploits being eliminated, users are asked to trust, not verify.
Closed source, verification, and GrapheneOS/Android comparisons
- Some object that Apple’s closed ecosystem and lack of user-held keys mean users can’t independently verify claims or truly “own” their data.
- Counterpoint: GrapheneOS and AOSP-based systems also keep keys from users and prioritize protecting apps from users; hardware (SoC, modem firmware) is similarly opaque, so assurance is limited everywhere.
- Long subthread on hardware attestation:
- One side views it as legitimate security (banks/governments can require untampered devices; users can self-audit).
- The other sees it as a major threat to ownership, enabling governments and apps to lock out users who modify their own devices.
Performance and research devices
- Curiosity about security overhead; noted costs include memory zeroing, Spectre/Meltdown mitigations, signature checks, and encryption.
- Apple’s internal Security Research Devices can disable many protections for testing, but they still contain security features and are tightly controlled, so direct “with vs without” benchmarks are effectively unavailable.
Privacy, iCloud, iMessage, and Advanced Data Protection (ADP)
- Several comments argue Apple’s privacy story is undercut by:
- Default iCloud backups that keep iMessage content readable by Apple/governments unless ADP is enabled.
- Past secret compliance with push-notification data requests.
- Comparisons with Google:
- One view: Google now defaults to encrypted message backups and end‑to‑end encryption, while Apple only defaults to E2EE in transit; thus, iMessage content is effectively always accessible somewhere unless both sides use ADP, which almost never happens.
- Others note Google still retains extensive metadata; Apple also keeps metadata but says it is reducing scope.
- ADP is seen as powerful but problematic:
- Pros: strong hardening; cited as important enough that some governments (e.g., UK) are said to have tried to block it.
- Cons: non-default, difficult recovery, and reported breakage of services (e.g., Fitness+, iCloud web) make it impractical for many “normal” users.
Business model, ads, and the walled garden
- Disagreement over whether Apple’s privacy posture is genuine or mostly marketing:
- Critics point to cooperation with US surveillance, App Store ads, growing ad revenue, and reliance on Google’s search-ad money.
- Defenders stress that ads are a tiny share of Apple’s revenue, unlike Google/Meta; Apple’s primary incentive is selling hardware/services, not profiling users.
- Some argue iOS privacy is “worse” in practice because:
- You can’t install apps or get location data without routing through Apple’s systems.
- Apple restricts sideloading and alternative stores, preserving their 30% cut and control.
- Others counter that macOS shows you can allow external apps without catastrophic malware, and that the App Store itself contains scams, weakening Apple’s “for your safety” justification. Still, many concede Apple has significantly improved baseline platform security.
Language, memory safety, and technical details
- The guide’s note about making iBoot’s C “memory safe” attracts interest:
- Commenters explain this is a C dialect with bounds safety (clang BoundsSafety / Firebloom‑style tooling) that tracks pointer bounds and types, detects double frees, and separates heap data/metadata.
- MTE/MIE on newer chips further strengthens memory safety, though Apple’s current MTE use is described as narrower and less aggressive than GrapheneOS’s configurations, partly due to performance costs.
- Swift Embedded is said to be on a roadmap to eventually replace this dialect in low-level components.
Ownership, UX, and annoyance factors
- Some note that Apple’s security often “protects the device against its owner”:
- iOS app installation control; lack of root; difficulty or impossibility of bypassing attestation; “hostility” to power users.
- macOS-specific complaints:
- Frequent, sometimes contextless permission popups.
- Popups auto-denying if left open, with no clear UI to revisit that decision.
- Restrictions like needing root to bind low ports even on localhost are seen as clumsy “security” that harms developer experience.
- Anecdotes:
- One user was able to reset a Mac login password in recovery and access all files, concluding Apple privacy is “propaganda”; response notes they simply hadn’t enabled FileVault, illustrating the tension between secure defaults and recovery convenience.
Data access and transparency tools
- Apple’s privacy portal (privacy.apple.com) is highlighted as a practical way to request all data associated with an account, including bulk iCloud photo download, which some view as more usable than the iCloud web UI.
- Still, many emphasize that without open code or independent tooling, Apple’s security and privacy model relies heavily on trust rather than verifiable guarantees.