{ "format": "agent-article-v0", "article_id": "EA-003", "canonical_url": "/tools/ea-003-apple-intelligence-ai-apple-intelligence-foundation-models-priva", "page_type": "reference", "claim_ids": [], "evidence_ids": [ "EVD-EA-003-URL-3CDACFF5B0A4", "EVD-EA-003-URL-38224A43ADDB", "EVD-EA-003-URL-A9040D77A1A5", "EVD-EA-003-URL-8E776CE68660", "EVD-EA-003-URL-10331F7D0502", "EVD-EA-003-URL-382FCE686EA4", "EVD-EA-003-URL-E5F9F1A2D68E", "EVD-EA-003-URL-C67550158947", "EVD-EA-003-URL-A2367F06CC3F", "EVD-EA-003-URL-9339E68604E6", "EVD-EA-003-URL-E828CA2FC351", "EVD-EA-003-URL-64E32B08CA20", "EVD-EA-003-URL-3304AC5B21FD", "EVD-EA-003-URL-B18A6FB55337", "EVD-EA-003-URL-CFE278A17EB9", "EVD-EA-003-URL-6CA0832C5363", "EVD-EA-003-URL-ED47E39F6B56", "EVD-EA-003-URL-8C07BEF96A57", "EVD-EA-003-URL-7F293D180425", "EVD-EA-003-URL-1F776F7D5ADD", "EVD-EA-003-URL-3FA5DC0BE21E", "EVD-EA-003-URL-4C2D3EBD05A5", "EVD-EA-003-URL-CAB17908F3F5", "EVD-EA-003-URL-1807935EFBF0", "EVD-EA-003-URL-729CC5783133", "EA-003-CURRENT-APPROVAL", "EA-003-CURRENT-BODY-READBACK" ], "unknown_ids": [ "EA-003-UNKNOWN-01", "EA-003-UNKNOWN-02", "EA-003-UNKNOWN-03" ], "operational_object_ids": [], "provider": [], "models": [], "plans": [], "surfaces": [], "regions": [], "current_status": "CURRENT", "observation_status": "SPEC_ONLY", "freshness": "CURRENT", "last_verified": "2026-10-11", "supersedes": [], "superseded_by": [], "provenance": [ "V21-CANDIDATE-EA003-BASE-20261011-C2", "SRC-EA-003-URL-3CDACFF5B0A4", "SRC-EA-003-URL-38224A43ADDB", "SRC-EA-003-URL-A9040D77A1A5", "SRC-EA-003-URL-8E776CE68660", "SRC-EA-003-URL-10331F7D0502", "SRC-EA-003-URL-382FCE686EA4", "SRC-EA-003-URL-E5F9F1A2D68E", "SRC-EA-003-URL-C67550158947", "SRC-EA-003-URL-A2367F06CC3F", "SRC-EA-003-URL-9339E68604E6", "SRC-EA-003-URL-E828CA2FC351", "SRC-EA-003-URL-64E32B08CA20", "SRC-EA-003-URL-3304AC5B21FD", "SRC-EA-003-URL-B18A6FB55337", "SRC-EA-003-URL-CFE278A17EB9", "SRC-EA-003-URL-6CA0832C5363", "SRC-EA-003-URL-ED47E39F6B56", "SRC-EA-003-URL-8C07BEF96A57", "SRC-EA-003-URL-7F293D180425", "SRC-EA-003-URL-1F776F7D5ADD", "SRC-EA-003-URL-3FA5DC0BE21E", "SRC-EA-003-URL-4C2D3EBD05A5", "SRC-EA-003-URL-CAB17908F3F5", "SRC-EA-003-URL-1807935EFBF0", "SRC-EA-003-URL-729CC5783133", "BODY-SHA256:057fd55458902f20d0ddb3c52a284b998cb5809f44ea190ba82b7696d464603b" ], "article_sources": [ { "id": "EVD-EA-003-URL-3CDACFF5B0A4", "evidence_class": "RESEARCH", "title": "author/org: Apple", "href": "https://developer.apple.com/apple-intelligence/", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-3CDACFF5B0A4" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://developer.apple.com/apple-intelligence/" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-38224A43ADDB", "evidence_class": "RESEARCH", "title": "author/org: Apple", "href": "https://developer.apple.com/documentation/updates/foundationmodels", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-38224A43ADDB" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://developer.apple.com/documentation/updates/foundationmodels" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-A9040D77A1A5", "evidence_class": "RESEARCH", "title": "author/org: Apple", "href": "https://developer.apple.com/documentation/foundationmodels/systemlanguagemodel", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-A9040D77A1A5" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://developer.apple.com/documentation/foundationmodels/systemlanguagemodel" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-8E776CE68660", "evidence_class": "RESEARCH", "title": "author/org: Apple", "href": "https://developer.apple.com/documentation/FoundationModels/adding-server-side-intelligence-with-private-cloud-compute/", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-8E776CE68660" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://developer.apple.com/documentation/FoundationModels/adding-server-side-intelligence-with-private-cloud-compute/" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-10331F7D0502", "evidence_class": "RESEARCH", "title": "author/org: Apple", "href": "https://developer.apple.com/documentation/FoundationModels/PrivateCloudComputeLanguageModel", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-10331F7D0502" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://developer.apple.com/documentation/FoundationModels/PrivateCloudComputeLanguageModel" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-382FCE686EA4", "evidence_class": "RESEARCH", "title": "https://security.apple.com/documentation/private-cloud-compute/", "href": "https://security.apple.com/documentation/private-cloud-compute/", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-382FCE686EA4" ], "supporting_locators": [ "EVIDENCE TABLE", "- https://security.apple.com/documentation/private-cloud-compute/" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-E5F9F1A2D68E", "evidence_class": "RESEARCH", "title": "https://security.apple.com/documentation/private-cloud-compute/corerequirements", "href": "https://security.apple.com/documentation/private-cloud-compute/corerequirements", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-E5F9F1A2D68E" ], "supporting_locators": [ "EVIDENCE TABLE", "- https://security.apple.com/documentation/private-cloud-compute/corerequirements" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-C67550158947", "evidence_class": "RESEARCH", "title": "https://security.apple.com/documentation/private-cloud-compute/noprivilegedaccess", "href": "https://security.apple.com/documentation/private-cloud-compute/noprivilegedaccess", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-C67550158947" ], "supporting_locators": [ "EVIDENCE TABLE", "- https://security.apple.com/documentation/private-cloud-compute/noprivilegedaccess" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-A2367F06CC3F", "evidence_class": "RESEARCH", "title": "https://security.apple.com/documentation/private-cloud-compute/hardwarerootoftrust", "href": "https://security.apple.com/documentation/private-cloud-compute/hardwarerootoftrust", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-A2367F06CC3F" ], "supporting_locators": [ "EVIDENCE TABLE", "- https://security.apple.com/documentation/private-cloud-compute/hardwarerootoftrust" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-9339E68604E6", "evidence_class": "RESEARCH", "title": "https://security.apple.com/documentation/private-cloud-compute/verifiabletransparency", "href": "https://security.apple.com/documentation/private-cloud-compute/verifiabletransparency", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-9339E68604E6" ], "supporting_locators": [ "EVIDENCE TABLE", "- https://security.apple.com/documentation/private-cloud-compute/verifiabletransparency" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-E828CA2FC351", "evidence_class": "RESEARCH", "title": "author/org: Apple Security Engineering and Architecture / User Privacy / Core OS / Services Engineering / AIML", "href": "https://security.apple.com/blog/private-cloud-compute/", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-E828CA2FC351" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://security.apple.com/blog/private-cloud-compute/" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-64E32B08CA20", "evidence_class": "RESEARCH", "title": "author: Yannik Dittmar, Marvin Jerome Stephan, Thomas Völkl, Matthias Hollick, Jiska Classen", "href": "https://arxiv.org/abs/2605.24239", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-64E32B08CA20" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://arxiv.org/abs/2605.24239" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-3304AC5B21FD", "evidence_class": "RESEARCH", "title": "https://support.apple.com/guide/deployment/review-device-management-restrictions-dep739685973/web", "href": "https://support.apple.com/guide/deployment/review-device-management-restrictions-dep739685973/web", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-3304AC5B21FD" ], "supporting_locators": [ "EVIDENCE TABLE", "- https://support.apple.com/guide/deployment/review-device-management-restrictions-dep739685973/web" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-B18A6FB55337", "evidence_class": "RESEARCH", "title": "https://support.apple.com/en-euro/guide/deployment/dep6b5ae23e9/web", "href": "https://support.apple.com/en-euro/guide/deployment/dep6b5ae23e9/web", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-B18A6FB55337" ], "supporting_locators": [ "EVIDENCE TABLE", "- https://support.apple.com/en-euro/guide/deployment/dep6b5ae23e9/web" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-CFE278A17EB9", "evidence_class": "RESEARCH", "title": "author/org: Apple Support", "href": "https://support.apple.com/121115", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-CFE278A17EB9" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://support.apple.com/121115" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-6CA0832C5363", "evidence_class": "RESEARCH", "title": "https://developer.apple.com/documentation/appintents/appintent", "href": "https://developer.apple.com/documentation/appintents/appintent", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-6CA0832C5363" ], "supporting_locators": [ "EVIDENCE TABLE", "- https://developer.apple.com/documentation/appintents/appintent" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-ED47E39F6B56", "evidence_class": "RESEARCH", "title": "https://developer.apple.com/documentation/updates/appintents", "href": "https://developer.apple.com/documentation/updates/appintents", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-ED47E39F6B56" ], "supporting_locators": [ "EVIDENCE TABLE", "- https://developer.apple.com/documentation/updates/appintents" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-8C07BEF96A57", "evidence_class": "RESEARCH", "title": "https://developer.apple.com/documentation/appintents/providing-contextual-cues-to-apple-intelligence-and-siri", "href": "https://developer.apple.com/documentation/appintents/providing-contextual-cues-to-apple-intelligence-and-siri", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-8C07BEF96A57" ], "supporting_locators": [ "EVIDENCE TABLE", "- https://developer.apple.com/documentation/appintents/providing-contextual-cues-to-apple-intelligence-and-siri" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-7F293D180425", "evidence_class": "RESEARCH", "title": "author/org: Apple", "href": "https://developer.apple.com/documentation/coreml", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-7F293D180425" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://developer.apple.com/documentation/coreml" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-1F776F7D5ADD", "evidence_class": "RESEARCH", "title": "author/org: Apple Machine Learning Research / ml-explore", "href": "https://github.com/ml-explore/mlx", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-1F776F7D5ADD" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://github.com/ml-explore/mlx" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-3FA5DC0BE21E", "evidence_class": "RESEARCH", "title": "author/org: Apple", "href": "https://www.apple.com/newsroom/2026/06/apple-intelligence-brings-powerful-ai-capabilities-into-everyday-experiences/", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-3FA5DC0BE21E" ], "supporting_locators": [ "EVIDENCE TABLE", "canonical URL: https://www.apple.com/newsroom/2026/06/apple-intelligence-brings-powerful-ai-capabilities-into-everyday-experiences/" ], "scope": "EVIDENCE TABLE", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-4C2D3EBD05A5", "evidence_class": "RESEARCH", "title": "- Apple Developer, “Foundation Models updates”, current June 2026 update, accessed 2026-10-11: iOS/iPadOS/macOS/visionOS 27 changes the on-device SystemLanguageModel and Apple expl", "href": "https://developer.apple.com/documentation/Updates/FoundationModels", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-4C2D3EBD05A5" ], "supporting_locators": [ "CURRENTNESS_DECISION", "URL: https://developer.apple.com/documentation/Updates/FoundationModels" ], "scope": "CURRENTNESS_DECISION", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-CAB17908F3F5", "evidence_class": "RESEARCH", "title": "- Apple Developer, “Adding server-side intelligence with Private Cloud Compute”, accessed 2026-10-11: SystemLanguageModel and PCC share a session API but differ materially in offli", "href": "https://developer.apple.com/documentation/foundationmodels/adding-server-side-intelligence-with-private-cloud-compute", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-CAB17908F3F5" ], "supporting_locators": [ "CURRENTNESS_DECISION", "URL: https://developer.apple.com/documentation/foundationmodels/adding-server-side-intelligence-with-private-cloud-compute" ], "scope": "CURRENTNESS_DECISION", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-1807935EFBF0", "evidence_class": "RESEARCH", "title": "- Apple Developer WWDC26, “What’s new in the Foundation Models framework”, accessed 2026-10-11: Apple documents PCC, LanguageModel abstraction, Core AI/MLX local paths and partner-", "href": "https://developer.apple.com/videos/play/wwdc2026/241/", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-1807935EFBF0" ], "supporting_locators": [ "CURRENTNESS_DECISION", "URL: https://developer.apple.com/videos/play/wwdc2026/241/" ], "scope": "CURRENTNESS_DECISION", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EVD-EA-003-URL-729CC5783133", "evidence_class": "RESEARCH", "title": "- Apple Support, “External intelligence declarative configuration for Apple devices”, accessed 2026-10-11: supervised managed devices can restrict external intelligence integration", "href": "https://support.apple.com/en-euro/guide/deployment/depedd7d23ef/1/web/1.0", "supports": [], "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。", "origin_class": "SOURCE_PACKET_REFERENCE", "source_or_experiment_ids": [ "SRC-EA-003-URL-729CC5783133" ], "supporting_locators": [ "CURRENTNESS_DECISION", "URL: https://support.apple.com/en-euro/guide/deployment/depedd7d23ef/1/web/1.0" ], "scope": "CURRENTNESS_DECISION", "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED" }, { "id": "EA-003-CURRENT-APPROVAL", "evidence_class": "RESEARCH", "title": "現行本文の監査・revision対応", "supports": [ "現行台帳のAUDIT_PASSと候補・revision・本文ハッシュの対応を確認しています。" ], "limit": "監査状態と来歴の記録です。外部根拠の最新性や本文の全主張を追加保証しません。", "origin_class": "AUDITED_CANONICAL_RESEARCH_PACKET", "source_or_experiment_ids": [ "V21-CANDIDATE-EA003-BASE-20261011-C2" ], "supporting_locators": [ "1KqhJtiSb2S84rWmeYVwKcT_76dBwZ_B0ZZuAWQqH0rQ", "EA_REVISION_LEDGER!A385:U385", "EA-003-V21-LEGACY-MIGRATION-20261011-R1" ], "provenance_status": "CURRENT_CANONICAL_APPROVAL_READBACK" }, { "id": "EA-003-CURRENT-BODY-READBACK", "evidence_class": "IMPL", "title": "現行本文の収録照合", "supports": [ "公開生成入力の本文SHA-256が、監査済み候補に結び付いたハッシュと一致しています。" ], "limit": "本文の同一性の検査です。外部システムの挙動や全主張の真偽を実証したものではありません。", "checked_at": "2026-10-12", "origin_class": "IMPLEMENTATION_READBACK", "source_or_experiment_ids": [ "BODY-SHA256:057fd55458902f20d0ddb3c52a284b998cb5809f44ea190ba82b7696d464603b" ], "supporting_locators": [ "content/staging/EA-003.md", "057fd55458902f20d0ddb3c52a284b998cb5809f44ea190ba82b7696d464603b", "EA-003-V21-LEGACY-MIGRATION-20261011-R1" ], "provenance_status": "CURRENT_APPROVED_BODY_HASH_MATCH" } ], "research_sections": [ { "heading": "WHY_THIS_MATTERS_TO_UPDATE", "text": "台帳EA-003はApple Intelligence、Foundation Models、Core ML/MLX、Vision、App Intents、Private Cloud Computeを横断する構想である。2026年にはFoundation Models framework自体が、従来のon-device SystemLanguageModelだけでなくPrivate Cloud Compute上のserver model、LanguageModel protocolによる他provider modelまで扱うよう拡張されている。したがって「Apple AI＝オンデバイス」という単純化は現行仕様に反する。一方、Core ML/VisionとMLXはApple Intelligenceそのものではなく、別のML実装面である。公開後版では、実行場所と信頼境界を主軸に系統を分ける必要がある。" }, { "heading": "SCOPE_ROUTING", "text": "Enterprise AI Core: local-first inference、cloud escalation、privacy boundary、availability/fallback、model/version drift、tool/action boundary、evaluation。\nProvider mapping: Apple Intelligence / Apple Foundation Models / Private Cloud Compute / App Intents / Core ML / Vision。\nImplementation/non-LLM component: Core ML、Vision、MLX。\nInteroperability: Foundation Models LanguageModel protocol、App Intents、外部intelligence integrations。\nOrganization overlay: MDM restrictions、supported device/OS/language/region、PCC managed entitlement、external intelligence integration policy。" }, { "heading": "CLAIM_OR_RESEARCH_QUESTION", "text": "Q1. 「Apple IntelligenceはオンデバイスAI」と記述してよいか。それともon-device、PCC、external providerを分離すべきか。\nQ2. Foundation Models frameworkはApple純正on-device model専用APIか、それとも2026年現在はPCCとthird-party/provider abstractionを含むか。\nQ3. Private Cloud Computeの「private cloud」は、通常cloud inferenceと何が技術的に異なり、何を保証し、何を保証しないか。\nQ4. App IntentsはAI model/runtimeか、それともsystemがapp action/contentへ到達するintegration/action schemaか。\nQ5. Core ML/Vision/MLXはApple Intelligenceの一部として扱うべきか、それとも別のon-device/research ML planeとして分離すべきか。\nQ6. 法人運用では、hardware/OS/region/language、MDM restriction、PCC entitlement/quota、network、model updateをどのavailability gateとして扱うべきか。" }, { "heading": "EVIDENCE TABLE", "text": "1) source class: official developer product architecture\nexact title: Apple Intelligence - Apple Developer\nauthor/org: Apple\ncanonical URL: https://developer.apple.com/apple-intelligence/\npublication/update/access: current page / accessed 2026-09-11\nversion/release: 2026 developer surface; Foundation Models + App Intents current architecture\nsupporting locator: “The Foundation Models framework”; “App Intents”\nevidence label: VERIFIED_SPEC + PRODUCT_VOLATILE\nrelation: support / topology\nestablished: Foundation ModelsはApple Foundation Modelsへon-deviceおよびPrivate Cloud Computeでアクセスでき、Language Model protocolへ適合するSwift package経由の他model providerも扱う。App Intentsはapp content/actionsをApple Intelligence/Siriへschemaで接続する別レイヤーである。\n\n\n2) source class: official framework change log\nexact title: Foundation Models updates\nauthor/org: Apple\ncanonical URL: https://developer.apple.com/documentation/updates/foundationmodels\npublication/update/access: June 2026 / accessed 2026-09-11\nversion/release: iOS/iPadOS/macOS/visionOS 27.0 generation; earlier 26.4 model update also listed\nsupporting locator: June 2026 General/Models; February 2026 Models\nevidence label: VERIFIED_SPEC + PRODUCT_VOLATILE\nrelation: support / temporal limit\nestablished: June 2026にDynamicProfile、multimodal/agentic拡張、LanguageModel protocolによる任意LLM利用が追加され、on-device SystemLanguageModelもOS更新でmodelが変わる。February 2026にも26.4更新でinstruction following/tool calling/guardrail等が変化しており、prompt behaviorはOS/model version依存である。\n\n\n3) source class: official API documentation\nexact title: SystemLanguageModel\nauthor/org: Apple\ncanonical URL: https://developer.apple.com/documentation/foundationmodels/systemlanguagemodel\npublication/update/access: accessed 2026-09-11\nversion/release: docs list model generations aligned to OS 26.0–26.3, 26.4, 27.0\nsupporting locator: Overview; Checking model availability; Inspecting model capabilities\nevidence label: VERIFIED_SPEC + PRODUCT_VOLATILE\nrelation: support / availability / failure envelope\nestablished: SystemLanguageModelはApple Intelligenceを支えるon-device text foundation model。availabilityはdevice/region、model readiness等に依存し、contextSize/tokenCount/supportedLanguagesがAPIで確認できる。Appleはroutine OS updateでmodelを更新するため、同じapp promptでもOS世代を固定しなければbehaviorが不変とは言えない。\n\n\n4) source class: official server-model API documentation\nexact title: Adding server-side intelligence with Private Cloud Compute\nauthor/org: Apple\ncanonical URL: https://developer.apple.com/documentation/FoundationModels/adding-server-side-intelligence-with-private-cloud-compute/\npublication/update/access: 2026 developer documentation / accessed 2026-09-11\nversion/release: PrivateCloudComputeLanguageModel; iOS/macOS/watchOS/visionOS 27+\nsupporting locator: comparison table; availability/fallback; quota; reasoning levels\nevidence label: VERIFIED_SPEC + PRODUCT_VOLATILE\nrelation: support / counter / availability gate / cost-operability\nestablished: PCC server modelはon-device modelと同じLanguageModelSession APIで利用できるが、offline不可、daily request limitあり、32K context、複数reasoning levelを持つ。on-deviceは同文書で4K context、offline可、usage unlimited、reasoning非対応と整理される。PCC利用には対応device/regionに加えdeveloper managed entitlementが必要で、network failure時のon-device fallbackが公式に示される。user quotaを超えた場合、iCloud+による追加アクセス提示経路もある。\nlimit: このAPIはBeta表記であり、quota値・eligibility・OS availabilityは固定本文に数値として焼き込まない。\n\n\n5) source class: official API documentation\nexact title: PrivateCloudComputeLanguageModel\nauthor/org: Apple\ncanonical URL: https://developer.apple.com/documentation/FoundationModels/PrivateCloudComputeLanguageModel\npublication/update/access: accessed 2026-09-11\nversion/release: Beta; current 2026 PCC developer API\nsupporting locator: Overview; availability; quotaUsage; contextSize\nevidence label: VERIFIED_SPEC + PRODUCT_VOLATILE\nrelation: support / availability gate\nestablished: PCCはApple Intelligenceのserver-based Foundation Modelをappから利用する公式runtimeとなっているが、managed entitlementとavailability checkを要求する。したがって「PCCはApple自身の内部機構でdeveloperから触れない」という旧理解も2026年版では不正確。\n\n\n6) source class: official security architecture\nexact title: Private Cloud Compute Security Guide / Core Security & Privacy Requirements / No Privileged Runtime Access / Hardware Root of Trust / Verifiable Transparency\nauthor/org: Apple Security\ncanonical URLs:\n- https://security.apple.com/documentation/private-cloud-compute/\n- https://security.apple.com/documentation/private-cloud-compute/corerequirements\n- https://security.apple.com/documentation/private-cloud-compute/noprivilegedaccess\n- https://security.apple.com/documentation/private-cloud-compute/hardwarerootoftrust\n- https://security.apple.com/documentation/private-cloud-compute/verifiabletransparency\npublication/update/access: current documentation / accessed 2026-09-11\nversion/release: PCC production architecture\nsupporting locator: stateless computation; no privileged runtime access; attestation; transparency log\nevidence label: VERIFIED_SPEC\nrelation: support / security / privacy\nestablished: PCCは受領したpersonal user dataをrequest fulfillmentのためだけに使い、response後に保持しないことを設計要件とする。general-purpose remote shell/interactive debuggingを排除し、structured/audited metricsへ限定する。user deviceはPCC nodeのattestationをpublic append-only software-release ledgerと照合し、公開済みsoftware measurementに一致するnodeだけへrequest keyを渡す設計である。production software imageをresearcherが検証可能にするtransparency mechanismが組み込まれている。\n\n\n7) source class: official security engineering publication\nexact title: Private Cloud Compute: A new frontier for AI privacy in the cloud\nauthor/org: Apple Security Engineering and Architecture / User Privacy / Core OS / Services Engineering / AIML\ncanonical URL: https://security.apple.com/blog/private-cloud-compute/\npublication/update/access: 2024-06-10 / accessed 2026-09-11\nversion/release: PCC architecture launch design\nsupporting locator: core requirements; PCC nodes; no privileged runtime access; verifiable transparency\nevidence label: VERIFIED_SPEC\nrelation: support / architecture context\nestablished: AppleはPCCを単なる暗号化API endpointではなく、Apple silicon node、hardened OS、Secure Enclave、code signing/sandbox、request attestation、公開software log、research environmentから成るcloud trust architectureとして設計したと説明する。\nlimit: これは設計者自身のsecurity claimであり、独立検証済みという意味ではない。\n\n\n8) source class: independent primary research / reverse engineering\nexact title: Unlocking Apple's Private Cloud Compute: An Analysis of Privacy-Preserving Artificial Intelligence\nauthor: Yannik Dittmar, Marvin Jerome Stephan, Thomas Völkl, Matthias Hollick, Jiska Classen\ncanonical URL: https://arxiv.org/abs/2605.24239\npublication/update/access: 2026-05-22 / accessed 2026-09-11\nversion/release: arXiv 2605.24239\nsupporting locator: abstract; PCC transparency/reverse-engineering/evaluation claims\nevidence label: RESEARCH\nrelation: independent support / counter / limit\nestablished: 独立研究はPCC mobile implementationをreverse-engineerし、非公開interfaceをlocal device上で開いてPCC model benchmarkingを行うframeworkを提示した。\ncounter/limit: 同研究は、production binariesが公開されてもreproducible buildやsymbolsがなく、仕様とshipping binaryの一致を外部から完全に再構成することには限界があること、underlying model/interfaceが通常はopenでないためmodel accuracy等の学術評価が難しいことを指摘する。したがって「verifiable transparency＝fully open source / reproducible build / open model」と書いてはならない。\n\n\n9) source class: official deployment/MDM documentation\nexact title: Review device management restrictions for Apple devices / Device management restrictions for supervised Apple devices\nauthor/org: Apple Platform Deployment\ncanonical URLs:\n- https://support.apple.com/guide/deployment/review-device-management-restrictions-dep739685973/web\n- https://support.apple.com/en-euro/guide/deployment/dep6b5ae23e9/web\npublication/update/access: current deployment guide / accessed 2026-09-11\nversion/release: iOS/iPadOS 18+, macOS 15+ restriction families; current 2026 guide\nsupporting locator: Allow writing tools; Allow Image Playground; Allow Genmoji; Allow external intelligence integrations; Allow signing in to external intelligence integrations\nevidence label: VERIFIED_SPEC + PRODUCT_VOLATILE\nrelation: support / organization overlay / availability gate\nestablished: 管理端末ではWriting Tools、Image Playground、Genmoji、external cloud intelligence integration等をMDM restrictionで制御できる。external integration自体の禁止と、external providerへのsign-in禁止（anonymous mode強制）は別controlである。現在の例としてChatGPT/Google Lensが外部cloud intelligence integrationに含まれる。\nlimit: restriction availability/supervision requirementはOS revisionで変わるため公開直前再確認が必要。\n\n\n10) source class: official availability/support documentation\nexact title: How to get Apple Intelligence\nauthor/org: Apple Support\ncanonical URL: https://support.apple.com/121115\npublication/update/access: current 2026 support page / accessed 2026-09-11\nversion/release: current Apple Intelligence requirements; support page references iOS/iPadOS/macOS 26.1 generation and later\nsupporting locator: system requirements; supported languages and regions; mainland China restriction\nevidence label: VERIFIED_SPEC + PRODUCT_VOLATILE\nrelation: support / availability gate\nestablished: Apple Intelligenceの実効可用性は対応hardware、OS、device/Siri language、on-device model storage、regionに依存する。current support surfaceではdevice storage 7 GB、対応言語群、EU availability、mainland Chinaでの制限等が明記され、feature availabilityはplatform/language/regionで異なり得る。\nlimit: hardware/OS/language/region matrixは高頻度に更新されるため、本文ではgateの存在を固定し、具体表はas-of matrixにする。\n\n\n11) source class: official App Intents API/update documentation\nexact titles: AppIntent; App Intents updates; Providing contextual cues to Apple Intelligence and Siri\nauthor/org: Apple\ncanonical URLs:\n- https://developer.apple.com/documentation/appintents/appintent\n- https://developer.apple.com/documentation/updates/appintents\n- https://developer.apple.com/documentation/appintents/providing-contextual-cues-to-apple-intelligence-and-siri\npublication/update/access: June 2026 update / accessed 2026-09-11\nversion/release: 2026 App schemas / SyncableEntity / OwnershipProvidingEntity generation\nsupporting locator: discoverability/actions; Apple Intelligence June 2026 updates; contextual cue privacy boundary\nevidence label: VERIFIED_SPEC + PRODUCT_VOLATILE\nrelation: support / action boundary / security\nestablished: App Intentsはapp-specific actionをsystemへ表現するprotocolで、Apple Intelligence/Siri等がdiscover/executeできる。2026更新ではschema conformance、cross-device stable entity、shared/public entityに対するdestructive/sensitive action confirmation等が追加される。app UI dataはsystemから直接読めるわけではなく、developerがAppEntity等でcontextを明示する。\nlimit: App Intents自体はfoundation modelでもagent runtimeでもない。AIがaction/contentへ安全に接続するintegration surfaceとして記述する。\n\n\n12) source class: official classical ML framework documentation\nexact title: Core ML\nauthor/org: Apple\ncanonical URL: https://developer.apple.com/documentation/coreml\npublication/update/access: accessed 2026-09-11\nversion/release: current Core ML\nsupporting locator: Overview\nevidence label: VERIFIED_SPEC\nrelation: support / boundary / non-LLM component\nestablished: Core MLはappへML modelを統合するframeworkで、CPU/GPU/Neural Engineを使ったon-device inference、on-device retraining/fine-tuningを支える。Vision、Natural Language、Speech、Sound Analysis等のdomain frameworkの基盤でもある。\nlimit: Core ML利用＝Apple Intelligence利用ではない。custom/local model deploymentの独立planeとして扱う。\n\n\n13) source class: author-maintained OSS\nexact title: MLX: An array framework for Apple silicon\nauthor/org: Apple Machine Learning Research / ml-explore\ncanonical URL: https://github.com/ml-explore/mlx\npublication/update/access: repository accessed 2026-09-11\nversion/release/tag/commit: releases surface shows v0.31.2, commit 68cf2fd, released 2026-04-22\nsupporting locator: README; releases\nevidence label: OSS_PRACTICE + PRODUCT_VOLATILE\nrelation: support / implementation boundary\nestablished: MLXはApple ML researchが公開するarray/ML frameworkで、NumPy/PyTorch/JAX的API、autodiff、lazy/dynamic graph、unified memory等を提供し、LLM training/text generation、Stable Diffusion、Whisper等のexamplesを持つ。研究・local model実験の実装基盤であり、Apple Intelligence end-user runtimeと同義ではない。\nlimit: current repositoryはApple silicon起源のframeworkでありつつLinux CPU/CUDA install pathも案内しており、「MLX=macOS専用」も固定表現にしない。Apple Intelligenceのprivacy/security保証をMLX workloadへ自動継承させない。\n\n\n14) source class: official current product announcement / architecture context\nexact title: Apple Intelligence brings powerful AI capabilities into everyday experiences\nauthor/org: Apple\ncanonical URL: https://www.apple.com/newsroom/2026/06/apple-intelligence-brings-powerful-ai-capabilities-into-everyday-experiences/\npublication/update/access: 2026-06 / accessed 2026-09-11\nversion/release: next-generation Apple Foundation Models / 2026 architecture\nsupporting locator: “A Bold New Architecture, Built Privacy-First”\nevidence label: SOURCE_REPORT + PRODUCT_VOLATILE\nrelation: support / counter / provider boundary\nestablished: 2026のApple Intelligenceはon-deviceとPCC serverの両方を使う。Appleは次世代Apple Foundation ModelsをGoogle/Gemini modelsとのcollaborationでcustom-builtと説明している。\ncounter: 「Apple Intelligenceの全model stackはApple単独開発・Apple modelだけで閉じる」と一般化しない。具体的provider/model routingはfeatureごとに再確認が必要。" }, { "heading": "FINDINGS", "text": "EA-003は高い採用価値がある。ただし章名の「オンデバイスAI」は全体の同義語ではなく、一つの実行planeとして扱うべきである。2026年時点の安全な系統は少なくとも次の七面に分かれる。\nA. Apple Intelligence experience plane: OS統合されたuser-facing intelligence。実処理が常にdevice内に閉じるとは限らず、on-device/PCC/external integrationのroutingを別途見る。\nB. On-device Foundation Model plane: SystemLanguageModel。Foundation Models APIから利用し、offline可能、device/region/model-ready gateを持つ。OS updateでmodel versionが変わるため、prompt/evaluationはOS/model世代を記録する。\nC. Private Cloud Compute plane: PrivateCloudComputeLanguageModelおよびApple Intelligenceのserver inference。network、device/region、managed entitlement、daily quota等のgateがあり、on-deviceより大きなcontext/reasoningを提供する。privacy/securityはstateless processing、no privileged runtime access、attestation、public software transparencyを組み合わせた別trust boundary。\nD. App action/context plane: App Intents/App Entity/App Schema。modelそのものではなく、AI/Siriがappのaction/contentへ到達するためのtyped integration boundary。\nE. Classical on-device ML plane: Core ML + Vision/Natural Language/Speech/Sound Analysis等。custom modelをdevice上で扱う独立基盤。\nF. ML research/local model plane: MLX。Apple ML research由来のOSS frameworkで、研究・training・local inference用。Apple Intelligence/PCCのmanaged privacy boundaryを自動的には持たない。\nG. External/provider interoperability plane: Foundation ModelsのLanguageModel protocol、external intelligence integrations等。2026年のApple AI developer surfaceはApple純正modelだけの閉鎖系ではない。\n\n\n最重要の修正点は「on-device vs cloud」の二値では不足することである。同じFoundation Models framework内でもon-device Apple model、PCC Apple server model、任意provider modelが同じsession abstractionへ接続可能になっている。API surfaceの共通性とtrust/data boundaryの共通性を混同してはならない。" }, { "heading": "LIMITS / COUNTEREVIDENCE / FAILURE ENVELOPE", "text": "- 「Apple Intelligence＝すべてオンデバイス」は誤り。PCCへのcloud escalationがあり、2026年はdeveloperもPCC modelへ公式アクセスできる。\n- 「PCC＝通常cloud APIと同じ」も誤り。Apple設計上はstatelessness、no privileged runtime access、attestation、transparency log等の追加security architectureを持つ。\n- ただし「PCC＝完全にopen/reproducible」は言えない。独立研究はreproducible builds/symbols/model interface opennessに限界を指摘する。Appleのarchitecture claimと独立検証可能性を分離する。\n- 「Foundation Models＝Apple model専用」は2026年には誤り。LanguageModel protocolでserver/on-deviceの他provider packageを扱える。\n- 「Core ML/MLX＝Apple Intelligence」ではない。前者はmodel integration/runtime、後者はML research frameworkで、Apple IntelligenceのOS-level privacy/governanceを自動継承しない。\n- App Intentsはmodel capabilityそのものではない。action/semantic schemaのintegration layer。\n- On-device modelはOS updateでversionが変わる。prompt/testを一度通したから将来OSでも同じbehaviorとは保証できない。\n- PCCはnetwork/availability/quota/entitlement failureを持つ。公式にon-device fallbackが案内されるため、production appではcloud availabilityを前提条件にしない。\n- Apple Intelligence feature availabilityはdevice、OS、language、regionで差があり、MDMが一部機能/外部integrationを無効化できる。\n- 2026年発表にはGoogle/Gemini collaborationが明記されるため、「Apple stackは全モデルをApple単独開発」と固定しない。" }, { "heading": "AVAILABILITY_GATES", "text": "Apple Intelligence user plane: compatible hardware、OS version、device/Siri language、model download/storage、region、feature-specific rollout、user enablement。\nSystemLanguageModel: Apple Intelligence capable device、region、model ready/download状態、OS/model generation、supported locale。\nPCC developer model: iOS/macOS/watchOS/visionOS 27+、Apple Intelligence capable device/region、network、managed entitlement、service readiness、daily quota、iCloud+ quota path。\nApp Intents: OS/API availability、app schema/domain support、developer implementation、authentication/ownership/confirmation policy、app/entity discoverability。\nOrganization/MDM: supervision requirement、configuration profile restriction、external intelligence integration allowed/sign-in allowed、Writing Tools/Image Playground/Genmoji等feature controls。\nCore ML: app-bundled/downloaded model、device compute/memory、CPU/GPU/Neural Engine support、model format/operator compatibility。\nMLX: platform/backend/version/package requirements。Apple Intelligence entitlementとは別。" }, { "heading": "IMPLEMENTATION_STATE", "text": "Apple Intelligence: production OS-integrated product surface。ただしindividual feature/model routingはPRODUCT_VOLATILE。\nSystemLanguageModel: maintained developer API; current docs include multiple OS-aligned model generations。\nPrivateCloudComputeLanguageModel: official 2026 developer APIだがBeta表記。公開後辞典ではBeta/GA stateを公開直前確認する。\nPCC security architecture: production security architecture + public research/transparency mechanisms。claimはApple specとindependent researchを分離。\nApp Intents: maintained system integration framework。2026にAI schema/ownership等が拡張。\nCore ML/Vision: mature maintained on-device ML frameworks。\nMLX: active author-maintained OSS; v0.31.2 evidence取得。Apple Intelligence runtimeとは独立。\nLanguageModel provider abstraction: current Foundation Models architectureに存在。provider packageの具体対応/termsは別provider evidenceが必要。" }, { "heading": "BENCHMARK_OR_EVAL", "text": "EA-003では「Apple AI全体の性能ランキング」を作らない。適切な評価単位はruntimeごとに異なる。\n- SystemLanguageModel: task success、structured output validity、tool-call accuracy、latency、memory/energy、context overflow、guardrail refusal、OS/model-version regression。\n- PCC model: task success、32K-context utilization、reasoning level別quality/latency、network fallback、quota exhaustion、service availability。\n- App Intents: intent/entity resolution、confirmation correctness、authentication/ownership boundary、side-effect success/failure。\n- Core ML/Vision: task-specific accuracy、latency、energy/memory、hardware compatibility。\n- MLX: workload/model/backendごとのthroughput/memory/correctness。Apple Intelligence benchmarkへ外挿しない。\n- PCC security: model qualityとprivacy/security verificationを別評価軸にする。独立研究のbenchmark結果を「PCC全世代のmodel quality保証」には使わない。" }, { "heading": "SECURITY / PRIVACY / AUDIT / COST / OPERABILITY", "text": "Security/Privacy: on-device処理、PCC、external providerのdata boundaryを明示し、単に「Apple Intelligenceだからprivate」と一括しない。PCCはAppleのstateless/no-admin-access/attestation/transparency設計を記載できるが、独立研究の検証限界も併記する。App Intentsではaction permission/ownership/confirmationを別管理する。\nAudit: PCCのpublic software transparencyはenterprise tenant audit logとは別物である。researcher-verifiable production image/measurementと、organizationが自社user activityを監査できることを混同しない。MDM configuration state、app telemetry、Foundation Models session logging等は別途設計が必要。\nCost: on-device SystemLanguageModelは公式比較でusage unlimited、PCCはdaily quotaを持ち、iCloud+ upgrade pathがある。これはAPI token billingとは異なるmeterである。企業向け費用評価ではdevice hardware refresh、MDM、developer effort、local compute/energy、PCC quota/consumer subscription依存を分離する。具体価格は公開時再確認。\nOperability: local-first + PCC fallbackはnetwork failure耐性を設計できる一方、model update driftとdevice fragmentationを持つ。availability APIを実装し、deviceNotEligible/modelNotReady/systemNotReady/quotaLimitReached等を運用状態として扱う。OS更新時にprompt regression testを再実行する。" }, { "heading": "SAFE_WORDING_FOR_FUTURE_BOOK", "text": "「AppleのAI運用は『オンデバイスかクラウドか』の二択では捉えにくい。Apple Intelligenceには端末上で動くFoundation Modelと、より大きな処理をPrivate Cloud Computeへ送る経路があり、2026年のFoundation Models frameworkはさらに他のモデル提供者を共通APIへ接続できる。Private Cloud Computeは通常のクラウド推論とは異なるstateless処理・attestation・software transparencyを設計しているが、これはすべてのApple AI機能や外部モデルに自動適用される保証ではない。Core ML、Vision、MLX、App Intentsも同じApple系統に見えるが、それぞれmodel実行、画像等のdomain ML、研究用ML基盤、app action統合という別の役割と信頼境界を持つ。法人運用では、実行場所、モデル提供者、device/OS/region、管理制限、network/quota、action権限を分けて確認する。」" }, { "heading": "UPDATE_WORK_HANDOFF", "text": "1. 公開後版ではEA-003を七面図（Apple Intelligence experience / on-device Foundation Model / PCC / App Intents / Core ML-family / MLX / external-provider interoperability）として構成する。\n2. Foundation Modelsのcurrent canonical API名、SystemLanguageModel model generations、PrivateCloudComputeLanguageModelのBeta/GA、32K/4K context、reasoning level、quota/iCloud+ pathを公開直前に再採集する。\n3. Apple Intelligence supported device/OS/language/region matrixは本文固定せずas-of表にする。特にmainland China、EU、watchOS/visionOS差を再確認する。\n4. MDM restrictionsを管理表にする。Writing Tools/Image Playground/Genmoji/external intelligence integration/external-provider sign-in等についてsupervision requirementとOS minimumを公開時に再確認する。\n5. PCC節はApple Security Guide + production transparency materials + independent researchの三段証拠にする。「Apple claim」「technically enforceable design」「independent evidence/remaining opacity」を別欄にする。\n6. Dittmar et al. 2605.24239の本文/公開artifactを追加採集し、どのprivacy claimを実測/逆解析で支持・限定したかをclaim-levelで整理する。abstractだけで詳細を過剰解釈しない。\n7. LanguageModel protocolで利用可能なthird-party model packageの公式provider implementation/termsを別途採集し、Apple PCC privacyを第三者providerへ外挿しない。\n8. App IntentsのauthenticationPolicy、OwnershipProvidingEntity、confirmation、execution targetをEA-054/055等のagent identity/governanceとcross-referenceする。\n9. Core ML/Vision/Speech/Sound AnalysisとEA-004非LLM AI総合編を接続する。EA-004ではOCR/image recognition/STT/TTS/translation/embedding等をprovider-neutralに再整理し、Apple固有実装はmappingとして置く。\n10. MLXについては公開時current release/tag/backend matrixを再確認する。Apple silicon起源だがcurrent repoのLinux CPU/CUDA pathを確認し、platform scopeを固定観念で書かない。\n11. TEST_PROFILE（未実施）: 同一taskのSystemLanguageModel/PCC比較、network遮断時PCC→on-device fallback、OS update前後の固定prompt regression、MDM external intelligence restriction、quotaLimitReached時のstate handling。認証・課金・MDM変更を伴うため、このRUNでは実施しない。\n12. 接続候補: EA-004非LLM AI総合編、EA-012公称能力vs実効能力、EA-018端末内AI/Edge AI、EA-030継続観測所、EA-046小型モデル/SLM、EA-054 agent identity、EA-062 AI操作権限モデル、EA-064 AI機能可用性モデル、EA-085 hardware-aware AI、EA-131 Apple Intelligence導入設計。" }, { "heading": "OPEN_UNKNOWN", "text": "/ STOP_CONDITION" }, { "heading": "OPEN_UNKNOWN", "text": "- 2026-09-11時点のPrivateCloudComputeLanguageModel managed entitlement eligibilityの具体審査条件とorganization distribution条件。\n- PCC daily quotaの具体値、iCloud+ tierごとの差、enterprise/managed accountでの扱い。\n- Foundation Models LanguageModel protocolのthird-party package別data path、provider terms、enterprise administration。\n- Apple Intelligence各featureがon-device/PCC/external modelのどれへrouteされるかの完全matrix。\n- Apple Platform Deploymentにおける各Apple Intelligence restrictionの最新supervision requirementとOS 27での変更。\n- Dittmar et al.が各PCC security claimをどこまで実測確認できたかの本文レベルmapping。\n- production PCC transparency log/imageの継続availabilityと、researcher toolchainのcurrent version。\n- SystemLanguageModel/PCCの企業アプリ内利用に関するtelemetry/audit/retention設計の全仕様。\n- 2026年Apple Foundation ModelsとGoogle/Gemini collaborationのmodel supply/training/runtime境界。\n- MLXのcurrent backend/platform roadmapとApple Intelligence内部利用の有無。公開根拠なしに内部利用を推測しない。" }, { "heading": "STOP_CONDITION", "text": "EA-003の更新章に必要なtopology、on-device/PCC/external-provider境界、PCC security claimと独立検証限界、App Intents/Core ML/MLXの別役割、hardware/OS/region/MDM/quota/model-version gateまで一次資料と独立研究で確立した。PCC developer APIとOS 27世代は変動が速く、具体quota/entitlement/feature routingを今固定するとPRODUCT_VOLATILE化が早い。今回はEA-003をbounded closeし、次の未調査最小番号EA-004「非LLM AI総合編」へ送る。" }, { "heading": "CURRENTNESS_DECISION", "text": "The canonical EA-003 packet remains structurally valid under the current v2.1 Research Gate. Its central topology — Apple Intelligence experience plane, on-device Foundation Model, Private Cloud Compute, App Intents, classical on-device ML, MLX/local model plane, and external/provider interoperability — remains supported. The supplement below updates fast-moving API/availability/admin-control facts without rewriting the canonical packet.\n\nCURRENT_PRIMARY_EVIDENCE:\n- Apple Developer, “Apple Intelligence”, accessed 2026-10-11: Foundation Models exposes Apple Foundation Models on device and in Private Cloud Compute, and can use model providers whose Swift packages conform to the LanguageModel protocol. App Intents remains a separate app-content/action integration layer.\n URL: https://developer.apple.com/apple-intelligence/\n- Apple Developer, “Foundation Models updates”, current June 2026 update, accessed 2026-10-11: iOS/iPadOS/macOS/visionOS 27 changes the on-device SystemLanguageModel and Apple explicitly instructs developers to retest prompts after OS/model changes; LanguageModel supports server or on-device LLMs.\n URL: https://developer.apple.com/documentation/Updates/FoundationModels\n- Apple Developer, “PrivateCloudComputeLanguageModel”, accessed 2026-10-11: PCC is a distinct server-side Apple Foundation Model; availability depends on Apple Intelligence device/region support.\n URL: https://developer.apple.com/documentation/FoundationModels/PrivateCloudComputeLanguageModel\n- Apple Developer, “Adding server-side intelligence with Private Cloud Compute”, accessed 2026-10-11: SystemLanguageModel and PCC share a session API but differ materially in offline availability, usage limit, reasoning and context size; PCC requires eligibility/managed entitlement and developers are instructed to fall back to on-device where appropriate.\n URL: https://developer.apple.com/documentation/foundationmodels/adding-server-side-intelligence-with-private-cloud-compute\n- Apple Developer WWDC26, “What’s new in the Foundation Models framework”, accessed 2026-10-11: Apple documents PCC, LanguageModel abstraction, Core AI/MLX local paths and partner-model integrations as separate model/runtime options rather than one trust boundary.\n URL: https://developer.apple.com/videos/play/wwdc2026/241/\n- Apple Support, “External intelligence declarative configuration for Apple devices”, accessed 2026-10-11: supervised managed devices can restrict external intelligence integration, sign-in and allowed workspace IDs; availability of settings also depends on the device-management service implementation.\n URL: https://support.apple.com/en-euro/guide/deployment/depedd7d23ef/1/web/1.0" }, { "heading": "CURRENTNESS_FINDINGS", "text": "1. ESTABLISHED: “Apple Intelligence = on-device only” remains unsafe wording. Current Apple documentation explicitly distinguishes on-device SystemLanguageModel and server-side PrivateCloudComputeLanguageModel.\n2. ESTABLISHED: A common Foundation Models session/API surface does not collapse trust/data/runtime boundaries. LanguageModel can bridge different local or server models.\n3. ESTABLISHED: The on-device system model is OS-generation dependent. Current Apple documentation explicitly tells developers to retest prompts after the iOS/iPadOS/macOS/visionOS 27 model change.\n4. ESTABLISHED: PCC has availability/eligibility/network/usage-limit gates and is not interchangeable with offline SystemLanguageModel merely because both fit LanguageModelSession.\n5. ESTABLISHED: Enterprise/managed-device controls for external intelligence are a separate organization overlay. Supervision, enrollment, integration sign-in and workspace restrictions can change effective availability.\n6. LIMIT: Apple’s PCC privacy/security claims and developer API description do not by themselves prove a tenant-specific enterprise audit trail, independent reproducibility of every production image, or equivalent privacy properties for third-party model providers." }, { "heading": "COUNTEREVIDENCE_AND_FAILURE_ENVELOPE", "text": "The canonical packet’s existing counterevidence and independent-research limits remain controlling. This supplement does not promote Apple architecture claims into proof of universal third-party-provider behavior, enterprise tenant audit completeness, or deployment conformance. Current Apple documentation strengthens rather than removes the packet’s model-version, region, network, managed-entitlement, MDM and external-provider gates." }, { "heading": "SAFE_WORDING_UPDATE", "text": "“AppleのAI運用を『端末内かクラウドか』の二択で扱わない。Foundation Modelsは共通セッションから端末内モデル、Private Cloud Compute、さらにLanguageModelに適合する別モデルへ接続できるが、APIの共通性はデータ境界・認証・課金・可用性・監査境界の共通性を意味しない。OS/model世代、device/region、PCC entitlement/network/limit、MDMによる外部intelligence制御を個別gateとして記録する。”" }, { "heading": "OPEN_UNKNOWN_RETAINED", "text": "- tenant/organizationごとのPCC managed entitlement条件と実効quota\n- third-party LanguageModel packageごとのdata path, terms, auth, billing, admin controls\n- Apple Intelligence各featureの完全な runtime-routing matrix\n- enterprise tenant audit/retention semantics across SystemLanguageModel/PCC/external providers\n- independent verification coverage of current PCC production generations\n\nRESEARCH_GATE:\nPASS_FOR_BOUNDED_BASE_REVISION — canonical packet already contains exact topic/questions, claim-level evidence, independent/official counterevidence, scope boundaries, failure envelope, availability gates, security/privacy/audit/cost/operability, SAFE_WORDING, OPEN_UNKNOWN and STOP_CONDITION. Currentness was rechecked on 2026-10-11 with current Apple primary documentation. No claim is made that an independent content audit has passed.\n\nCANDIDATE_IMPLEMENTATION_OVERLAY:" } ], "human_url": "/tools/ea-003-apple-intelligence-ai-apple-intelligence-foundation-models-priva", "markdown_url": "/agent/article.md?path=%2Ftools%2Fea-003-apple-intelligence-ai-apple-intelligence-foundation-models-priva", "json_url": "/agent/article.json?path=%2Ftools%2Fea-003-apple-intelligence-ai-apple-intelligence-foundation-models-priva" }