Apple Intelligence Not Working? Fix Common Issues on iPhone & Mac

What Causes This Issue

Since its rollout, Apple Intelligence has exposed deep-seated architectural constraints across iOS, iPadOS, and macOS. Users in the Apple Support Community report widespread activation failures, silent Siri fallbacks, and model download stalls — symptoms of a platform reconciling decades of legacy code with new on-device ML pipelines. The core friction points include insufficient local storage for foundation models, outdated Secure Enclave firmware on pre-2023 devices, corrupted Neural Engine caches, and entitlement mismatches between system daemons and the new Intelligence Services framework. Network-level blocks on Apple’s Private Cloud Compute endpoints also trigger silent degradation to legacy Siri behavior.

Step-by-Step Fixes

1. Verify device eligibility and OS version. Open Settings > General > About. Confirm iPhone 15 Pro or later, iPad with M1 or later, or Mac with M1 or later. Ensure iOS 18.1 / iPadOS 18.1 / macOS Sequoia 15.1 or newer. Earlier builds lack the Intelligence Services entitlements.
2. Free at least 12 GB of local storage. The on-device foundation models require ~7 GB; temporary expansion during compilation needs headroom. Settings > General > iPhone Storage. Offload unused apps, delete large message attachments, or enable Optimize Photos.
3. Force re-download of intelligence models. Settings > Apple Intelligence & Siri > Foundation Models > Remove Downloads. Restart device. Return to same screen; models re-fetch over Wi-Fi. Watch for “Preparing” spinner — do not interrupt.
4. Reset Neural Engine cache. On macOS: Terminal > sudo kextunload -b com.apple.driver.AppleNeuralEngine then sudo kextload -b com.apple.driver.AppleNeuralEngine. On iOS/iPadOS: Settings > General > Transfer or Reset > Reset > Reset All Settings (preserves data, clears ML caches).
5. Sign out and back into Apple Account. Settings > [Your Name] > Sign Out. Choose to keep data on device. Sign in again. This refreshes Private Cloud Compute identity tokens that gate model personalization.
6. Check network restrictions. Corporate MDM, DNS filtering (e.g., NextDNS, Pi-hole), or VPN profiles may block *.apple-intelligence.com and *.pcc.apple.com. Temporarily disable to validate. If blocked, add allowlist entries for Apple’s 17.0.0.0/8 and 192.168.0.0/16 ranges used by PCC.
7. Update Secure Enclave firmware (Mac only). Apple Menu > System Settings > General > Software Update. Install any “Security Response” or firmware updates. M-series Macs require current SEP firmware for model encryption keys.

Additional Solutions

Disable and re-enable Apple Intelligence. Settings > Apple Intelligence & Siri > toggle off, wait 30 seconds, toggle on. This forces a clean handshake with intelligenceplatformd daemon. Users in the Apple Support Community report this resolves stale entitlement caches after major OS upgrades.

Run Apple Diagnostics. Shut down Mac. Power on holding D. On iPhone: Settings > Privacy & Security > Analytics & Improvements > Run Diagnostics. Hardware faults in the Neural Engine or unified memory controller manifest as silent Intelligence failures without kernel panics.

Delete corrupted Siri voice models. Settings > Siri > Siri Voice > select different voice, then reselect preferred. This flushes /var/mobile/Library/Assistant caches that conflict with new semantic indexing pipelines.

Check Screen Time restrictions. Settings > Screen Time > Content & Privacy Restrictions > Intelligence & Siri. Ensure “Allow Apple Intelligence” is enabled. MDM profiles can override this silently.

Perform clean install via IPSW (advanced). Backup via Finder/iTunes. Download latest IPSW for device. Option-click Restore in Finder. Select IPSW. This eliminates filesystem-level corruption in /System/Library/PrivateFrameworks/IntelligenceServices.framework that in-place updates miss.

When to Contact Apple Support

Open a support case if: Intelligence models fail to download after three attempts on different Wi-Fi networks; Neural Engine diagnostic returns error NEC-0x3F; device enters boot loop after Secure Enclave firmware update; Private Cloud Compute requests return HTTP 403 with valid Apple Account; or Intelligence features remain greyed out on eligible hardware running current OS with 15+ GB free storage. Provide sysdiagnose (Settings > Privacy > Analytics > Share) and note exact timestamp of failure.

FAQ

Why does Apple Intelligence work on my iPhone but not my Mac? Separate model downloads per device. Mac requires M1 or later; Intel Macs are unsupported. Verify both devices meet storage and OS minimums independently.

Can I use Apple Intelligence offline? Core text generation, summarization, and Genmoji run fully on-device. Writing Tools, Image Playground, and ChatGPT fallback require Private Cloud Compute connectivity.

Does Low Power Mode disable Apple Intelligence? Yes. On-device model inference is deferred when battery < 20% or Low Power Mode active. Charge above 20% and disable Low Power Mode to restore.

Why did my custom Siri shortcuts stop working after enabling Intelligence? App Intents framework migrated shortcut handling. Re-record shortcuts in Shortcuts app to register new intent definitions. Legacy INIntent shortcuts are deprecated.

Is there a way to downgrade Apple Intelligence models to save space? No. Models are version-locked to OS build. Downgrading OS is only path to smaller models, not recommended due to security fixes.