Meta Removes Muse Image Feature After Privacy and Likeness Concerns
Meta 因隱私與肖像權疑慮移除 Muse Image 功能
Introduction
Meta has stopped a specific function in its Muse Image AI model that allowed users to create images based on public Instagram accounts.
Meta 已停止其 Muse Image AI 模型中的一項特定功能，該功能原先允許使用者根據公開的 Instagram 帳號建立圖像。
Main Body
The feature was added to the Meta AI chatbot on Tuesday by Meta Superintelligence Labs. While the rest of the Muse Image tools, such as photo editing, are still available, the option to use public Instagram profiles for image generation was turned off on Friday. This happened after Meta used an 'opt-out' system, meaning all public adult accounts were automatically included unless the user manually chose to remove their data from the AI training process.
這項功能是由 Meta Superintelligence Labs 於週二加入 Meta AI 聊天機器人的。雖然 Muse Image 的其他工具（例如照片編輯）仍可使用，但利用公開 Instagram 個人檔案生成圖像的選項已於週五關閉。這是因為 Meta 採用了「選擇退出」（opt-out）機制，意味著除非使用者手動選擇將其數據從 AI 訓練過程中移除，否則所有公開的成年人帳號都會被自動納入。
Several professional organizations led the opposition to this feature. For example, SAG-AFTRA argued that failing to require an 'opt-in' agreement was a serious mistake, as it created risks regarding digital replicas made without permission. Similarly, the Creative Artists Agency (CAA) emphasized that using a person's image or creative work in AI models without clear consent is unacceptable. These views were supported by privacy groups and famous creators, who shared guides on how to disable the feature online.
數個專業組織帶領反對此功能。例如，SAG-AFTRA 主張未要求「選擇加入」（opt-in）協議是一個嚴重錯誤，因為這會造成在未經許可下製作數位複製品的風險。同樣地，創意藝術家代理公司 (CAA) 強調，在沒有明確同意的情況下將個人的圖像或創意作品用於 AI 模型是不可接受的。這些觀點得到了隱私團體和知名創作者的支持，他們在網上分享了如何關閉此功能的指南。
This situation is part of a larger conflict between AI developers and the entertainment industry. A similar event occurred in 2025 when OpenAI stopped Sora 2 due to arguments over trademarked characters. Furthermore, more public figures are now registering trademarks to protect their likenesses, while experts continue to criticize the environmental impact of the large data centers needed for AI.
此次情況是 AI 開發商與娛樂業之間更大衝突的一部分。2025 年曾發生類似事件，當時 OpenAI 因商標角色的爭議而停止了 Sora 2。此外，現在有更多公眾人物註冊商標以保護其肖像，而專家則持續批評 AI 所需的大型數據中心對環境造成的影響。
Conclusion
Meta has removed the controversial feature, stating that it failed to meet user expectations regarding control and creative use.
Meta 已移除該項爭議功能，並表示其未能滿足使用者對於控制權與創意使用的期望。
Vocabulary Learning
⚡ The 'Sophisticated Connector' Shift
To move from A2 to B2, you must stop using and, but, and so for every sentence. Look at how this text connects complex ideas. It uses Transition Markers to guide the reader through an argument.
🛠️ From Basic to B2
| A2 Way (Simple) | B2 Way (Advanced) | Effect |
|---|---|---|
| Also... | Furthermore... | Adds a stronger, more formal point. |
| Like... | Similarly... | Compares two professional opinions. |
| Because of this... | Due to... | Creates a tighter cause-effect link. |
🔍 Deep Dive: The Logic of "Furthermore"
In the text, the author doesn't just list facts. They build a case.
- Fact A: OpenAI stopped Sora 2.
- Fact B: Public figures are registering trademarks.
By using "Furthermore," the writer tells us: "I have already given you one example of this conflict; now I am adding an even more important piece of evidence."
💡 Pro-Tip: The "Opt-In/Opt-Out" Logic
Notice the phrase "meaning all public adult accounts were automatically included."
Instead of starting a new sentence ("This means that..."), the B2 writer uses a comma and the present participle (meaning). This turns a clunky explanation into a fluid, professional observation. This is a hallmark of B2 fluency: merging a result directly into the sentence that describes the action.