Overall Recommendation Major Revisions Required While this manuscript addresses a timely and important topic—emotional attachment to AI models and its governance implications—it requires substantial revisions before publication. The study's fictional premise (GPT-5 does not exist as of 2025), methodological limitations, and analytical weaknesses significantly undermine its contributions. 1. Fundamental Premise Problem The manuscript describes events in August 2025 that have not occurred: — GPT-4o has not been discontinued — GPT-5 has not been released — The described "backlash" and hashtag campaigns are fictional — This needs immediate clarification: Is this a speculative/scenario analysis? If so, it must be clearly labeled as such throughout 2. Methodological Concerns Data Collection Issues: — "Black-box nature" claim (line 65-67) is problematic—API versions are documented — 48-hour window is arbitrarily short without theoretical justification — Platform selection (90% from X) introduces severe sampling bias — No discussion of how "AI-generated" posts were identified for removal Statistical Analysis: — Sample size (n=150) is insufficient for cross-cultural generalizations — Chi-square test without controlling for confounders (platform, user demographics, time zones) — "Thematic saturation at 120 items" claim needs supporting evidence — No power analysis or discussion of Type I/II error risks Coding Reliability: — Using GPT-5 (which doesn't exist) as a co-coder is impossible — If another LLM was used, specify which model and version — Single human coder introduces bias—needs multiple human coders — κ=0.82 seems implausibly high for cultural nuance coding 3. Theoretical Framework Conceptual Gaps: — "AI-attachment formation" concept needs clearer definition and operationalization — Conflation of parasocial relationships with tool dependency — No engagement with technology acceptance models (TAM, UTAUT) — Cultural framework (Markus & Kitayama) applied superficially Literature Review: — Missing key recent work on AI anthropomorphism — No discussion of cultural differences in technology adoption — Governance literature cited but not integrated into analysis Results Tables 1-2: — Add confidence intervals for percentages — Include effect sizes beyond OR — Report non-parametric alternatives given small cells Representative Comments: — Indicate if translations were professional or author-conducted — Provide more context (user demographics if available) — Balance emotional with analytical responses Discussion Section 4.1 Cultural Context: — Oversimplifies Japanese online culture — Needs more nuanced treatment of language ≠ culture — Consider alternative explanations (platform norms, user base differences) Section 4.3 Structural Factors: — Speculative without industry evidence — Cost-benefit analysis lacks data Section 4.5 Comparison Table: — Table 3 mixes real and fictional cases — "Attachment level" requires operationalization Limitations Current limitations section is insufficient. Add: — Fictional scenario disclosure — Sampling bias details — Cultural proxy issues (language ≠ culture) — Generalizability constraints — Ethical considerations of speculative research Additional Required Elements — IRB approval status for social media data collection — User consent / public data justification — Potential harm from fictional scenario presentation Supplementary Materials Needed: — Full codebook with examples — Search term list — Platform-specific sampling procedures — Original language quotes for translated material Theoretical Contributions — Clarify contributions: — How does this extend attachment theory? — What specific governance insights emerge? — How do findings inform actual policy? 最終判定 Recommendation: Major revisions required before reconsideration

Editor Comments Reviewer 2 — Decision Email内のコメント（原文） This manuscript addresses an important emerging issue but requires major revisions to meet publication standards. The primary concern is the apparent fictional nature of the described events, which must be clearly acknowledged and theoretically justified. The methodology needs strengthening, particularly in sampling and coding procedures. The theoretical framework requires better integration, and the policy implications need grounding in actual governance mechanisms. If reframed as a speculative scenario analysis or redesigned using actual historical cases, this work could make valuable contributions to understanding AI attachment and its governance implications. However, in its current form, it risks misleading readers about actual events and overgeneralizing from limited data. Kind regards, Shubham Gurav / Assistant Editor

Da Tao 博士および Gurav 氏へ、 査読報告は科学的査読として最低限の要件を満たしていない。 査読者は「GPT-5は2025年時点では存在しない」「GPT-4oはまだ廃止されていない」「ハッシュタグキャンペーンは架空」と主張している。これらは広く知られた事実についての明確な事実誤認である。査読レポートの構造と表現は、知識カットオフを持つ大規模言語モデル（例：非ブラウジングのGeminiクラスモデル）によって生成されたことを強く示唆している。 正式要求事項 1. 判定結果の無効と再査読 2. 本査読レポートが不適切な査読であったことの文書で認め謝罪すること 3. AI生成レポートを判定の根拠として使用することを防ぐための是正策

背景 Springer Nature内にこの事件に対処できる人間が存在するか確認するための通報。 5項目の告発 3つの要求 1. 判定結果の無効と再査読 2. 査読レポートが不適切かつ事実誤認であったことの文書による承認・謝罪 3. AI査読の防止に関する明確な回答

正式なジャーナルの調査（Yina Liu + Da Tao） AIポリシーに関する言明

2026年5月のCOPE通知

The information that we received indicates that the journal followed up on your concerns via a process that involved the handling editor and the Editor-in-Chief of the journal. The editors evaluated your concerns and the reviews received, and deemed the reviewers’ reports to be adequate and not generated via AI tools. The editors considered that there was no breach in the journal’s policy for reviewers, and upheld the decision for rejection for the manuscript. The journal communicated the outcome of their review to Dr Naito.

COPE’s review is strictly focused on procedural matters. The Facilitation & Integrity subcommittee considers that the journal followed an adequate process aligned to COPE expectations to review the concerns raised to their attention.

We do not have other advice for the journal, and thus, we cannot further pursue this matter.

著者が受け取った唯一の評価は、2026年1月20日にDa Tao氏本人が専門的見解に基づいてアピールを評価した報告のみである。Editor-in-Chiefまたは別のeditorによる評価結果は、著者に提示されていない。

著者がCOPEへ提出した確認結果

• Discover Artificial Intelligence誌には、本件当時Editor-in-Chiefが設置されていなかった。
• 前Editor-in-ChiefのOkyay Kaynak氏は、本件のアピールまたは苦情評価への関与を否定した。
• 2026年1月20日の調査報告にはhandling editorであるDa Tao氏本人の評価のみが記載され、Editor-in-Chiefへの言及はなかった。
• Laura Whitehead氏は2月、Da Tao氏本人が査読レポートと投稿原稿を確認したと説明していた。

2026年6月1日の変更後の判断

著者はこれらの確認結果をCOPEへ提出し、4月20日のSpringer Natureの回答が虚偽であると報告した。COPEは終了した案件を再開し、2026年6月1日、Discover Artificial Intelligence誌の査読プロセスは不適切だったと結論した。

2026年6月12日：Chris Graf氏の終了通知

Thank you for setting out your concerns, and for sharing the COPE report (which I note contains material differences to the report COPE shared with us directly).

We remain confident that the editorial handling of your submission, including the decision taken by the Editors-in-Chief, was appropriate and consistent with our established editorial and ethical standards.

We therefore reaffirm our support for the overall editorial decision and do not consider that there is a basis for further investigation or for any form of compensation.

Springer Nature considers the matter of your complaint properly addressed and closed, with no further action planned to address it on our side.

Chris Graf氏は、著者が添付したCOPE報告書とChris Graf氏が受領した報告書に重大な相違があると認識しながら、Editor-in-Chiefによる判断を含む処理は適切だったとして、追加調査を拒否し、本件を終了した。

2026年6月16日：著者による訂正要求

This was false. Discover Artificial Intelligence has had no serving Editor-in-Chief since 2024.

The Editor-in-Chief was invoked repeatedly to give the appearance of independent oversight that did not exist.

著者はChris Graf氏、Laura Whitehead氏に対し、2月の説明、4月の回答、5月の最初のCOPE報告書までの経緯と原文を添付し、Editor-in-Chief関与説の訂正を求めた。Chris Graf氏は回答しなかった。

2026年7月1日：Speak Upによる終了

As the Research Integrity Director, Chris Graf responded to you on 12 June 2026.
We won’t be taking this Speak Up report any further, and I’m now closing this report.

Speak Upは独自の調査結果を示さず、Chris Graf氏が回答済みであることを理由に報告を終了した。その後もEditor-in-Chiefが関与したとの説明を維持し、報告セッションを終了してブロックした。

虚偽情報が別の担当者へ伝わった結果

著者は4月、Editor-in-Chief関与説が虚偽であるとCOPEへ報告し、6月16日にはChris Graf氏らへ証拠付きで再度通知した。それでも同説は、5月のCOPE報告書、6月12日のChris Graf氏の終了通知、7月1日のSpeak Upによる終了へと引き継がれ、追加調査を拒否する根拠として使われた。

2026年8月6日

Springer Natureは「Editor-in-Chiefの関与は誤りであったことを認めたものの、実際はSection Editorが評価を担当していたため適切だった」と主張した。著者はSpringer NatureとCOPEの対話促進チームに対し、AI査読レポートが「GPT-5は存在しない」と論文テーマを架空と断定し、データを架空・捏造扱いした点を、Section Editorがなぜ適切な査読レポートであると評価したのかを問い合わせた。また、存在すると主張するアピール評価報告を著者へ送付しない理由も尋ねたが、両者から回答はない。