Loading devserhii.com...
Last updated Aug 18, 2026
LocalMuse (“the App”) is an on-device AI image generation and editing app for iPhone and iPad. This Privacy Policy explains what information the App accesses, what remains on your device, and what may be transmitted when you download an AI model, enhance or translate a prompt, use voice input, or contact support.
In this policy, “we,” “us,” and “our” refer to the developer identified as the seller of LocalMuse on its App Store product page.
When you create or edit an image, enhance or translate a prompt, or enter a prompt using your voice, LocalMuse may process and store the following information locally:
Microphone audio used for voice dictation is processed as a live audio stream and is not saved by LocalMuse. Only the resulting text may be stored locally as part of your prompt or generation history.
Information stored by LocalMuse remains in the App’s private on-device storage unless you choose to save, share, or export it. We do not receive or remotely access this local content.
LocalMuse can optionally use a Qwen 3.5 2B 4-bit model to rewrite, expand, or refine a prompt before image generation.
The Qwen prompt model is not included in the App. If the model has not already been installed, LocalMuse displays the model provider, the technical data sent for the download, and an explicit permission choice before starting the download. If you do not choose Allow & Download, the model is not downloaded and the feature does not run using that model.
After the model has been downloaded, prompt enhancement is performed locally on your device. The original prompt and enhanced prompt are not sent to us, Hugging Face, the Qwen model provider, or a cloud inference service.
In Settings, you can choose between Apple Translation and the optional Qwen model as the prompt-translation engine.
When Apple Translation is selected, LocalMuse uses Apple’s Translation framework to detect the prompt language and translate the prompt into English. iOS or iPadOS may ask for permission to download system language data required for translation.
Translation performed through Apple’s Translation framework takes place using system services managed by Apple. LocalMuse does not send the original or translated prompt to us or to a cloud translation service.
Apple’s handling of system services and language-data downloads is governed by Apple’s own privacy practices. For more information, review the Apple Privacy Policy .
When Qwen is selected, the original prompt and its English translation are processed locally using the downloaded Qwen model. The prompt content is not sent to us, Hugging Face, the Qwen model provider, or a cloud inference service.
If the Qwen model has not already been installed, LocalMuse asks you to confirm its download before the Qwen enhancement or translation feature can be used.
LocalMuse allows you to enter an image prompt using your voice. When you start voice input, the App requests access to the microphone and passes live audio buffers to Apple’s Speech framework using SpeechAnalyzer and DictationTranscriber.
Speech recognition is performed on your iPhone or iPad. LocalMuse does not send microphone audio to us, Hugging Face, the Qwen model provider, or any cloud speech-recognition service.
LocalMuse does not record microphone audio to a file or store it in the App’s gallery, generation history, or local database.
When dictation ends, the App stops the audio and analysis session, discards its in-memory audio buffers, and releases its speech-processing resources. The operating system manages the memory used by the system speech-recognition model.
The recognized text is inserted into the prompt field. Like text entered manually, it may be stored locally as part of your prompt and generation history.
If you subsequently use prompt enhancement or translation, the recognized text is processed according to the local engine you selected in Settings. Voice-derived prompt text is not sent to a cloud prompt-processing service.
Before on-device speech recognition can be used for a particular language, iOS or iPadOS may need to download a compatible speech-recognition asset from Apple. This asset is not bundled with LocalMuse and is not stored in the App’s private container.
Speech-recognition assets are stored in system-managed storage and may be retained, updated, deallocated, removed, or downloaded again by the operating system. LocalMuse does not receive or remotely access these assets.
Apple manages system speech assets and their downloads under its own privacy practices. For more information, review the Apple Privacy Policy .
LocalMuse uses Apple’s system photo picker when you choose a source image. The App receives only the images you select and does not receive access to browse your entire Photos library.
When you choose to save a result, LocalMuse may request permission from iOS or iPadOS to add that image to Photos. Images saved to Photos remain there until you delete them and may be included in backups or synchronization services that you control.
If you use the iOS or iPadOS share sheet, the content you select is sent to the destination you choose. Apple and the selected destination process that content under their own terms and privacy policies.
Hidden creations may be protected using Apple’s Local Authentication framework. Face ID, Touch ID, or device-passcode verification is performed by the operating system.
LocalMuse receives only whether authentication succeeded or failed. The App does not receive or store biometric templates or biometric data.
Some LocalMuse features require downloading an AI model. This may include models used for image generation, image editing, upscaling, prompt enhancement, or prompt translation.
App-managed models, including the optional Qwen prompt model, are not downloaded automatically. Before any model-download request begins, LocalMuse displays an in-App notice that identifies the file provider, lists the technical data sent for the download, confirms which creative data is not sent, and asks for your explicit permission. If you choose Cancel instead of Allow & Download or Allow & Update, LocalMuse does not start the download or send a request to that provider.
Most model files are downloaded over HTTPS from Hugging Face, Inc. or its delivery infrastructure. One pinned runtime component used by one model may be downloaded over HTTPS from GitHub, Inc. or its delivery infrastructure. The in-App notice identifies the provider or providers for the selected model before you give permission.
Model-download requests do not include your prompts, enhanced prompts, translated prompts, dictated prompt text, microphone audio, source images, masks, generated images, Photos library, contact information, or biometric information.
To deliver the selected files, the provider receives your IP address, from which it may derive an approximate location, as well as the request time and requested model or file. The provider may also receive and retain operating-system information, device or user-agent information, and download activity. This technical information is used to deliver the files, operate and secure the download service, diagnose failures, and comply with applicable law. It is not sent for cloud AI inference, advertising, or cross-app tracking by LocalMuse.
We do not receive or control Hugging Face or GitHub server logs. Their handling and retention of technical request information is described in the Hugging Face Privacy Policy and the GitHub General Privacy Statement .
We confirm that every third party to which LocalMuse transmits data through the App must provide the same or equivalent protection described in this Privacy Policy and required by applicable law and the Apple App Review Guidelines. If we cannot ensure that level of protection for a provider, LocalMuse will not transmit data to that provider.
Your permission applies to the requests needed to download the model you selected. You may withhold permission by choosing Cancel, and you may stop an active download in the App. Withdrawing permission stops future requests for that download but cannot undo requests already delivered. Deleting a downloaded model removes the local model files but does not delete provider-controlled server logs; requests concerning those logs must be directed to the applicable provider.
If Apple Translation requires additional language data, the operating system may request permission to download that data from Apple. These system downloads are managed by iOS or iPadOS and are not controlled by LocalMuse.
Voice prompt input may require a compatible speech-recognition asset. These assets are downloaded from Apple and managed by iOS or iPadOS through Apple’s Speech framework and AssetInventory. They are separate from App-managed models downloaded from Hugging Face or GitHub.
Speech asset downloads do not include your microphone audio, recognized prompt text, source images, generated images, or gallery contents.
LocalMuse does not have an account system, advertising SDK, behavioral analytics SDK, cloud AI-generation backend, cloud prompt-processing backend, or cloud speech-recognition backend.
Through the App, we do not receive your original prompts, enhanced prompts, translated prompts, dictated prompt text, microphone audio, voice recordings, speech-recognition results, selected photos, masks, generated images, gallery contents, app preferences, biometric information, or downloaded models.
If you contact support, we receive only the information you choose to provide, such as your email address, message, screenshots, device model, operating-system version, App version, or diagnostic details.
Support information is used only to respond to your request, investigate technical problems, prevent abuse, protect legal rights, and comply with applicable law.
LocalMuse does not access the advertising identifier, display targeted advertising, or track users across apps or websites owned by other companies. We do not sell personal information or share it for cross-context behavioral advertising.
You control the content stored locally by LocalMuse. You may:
Microphone audio used for voice prompt input is not retained by LocalMuse and therefore does not require a separate deletion action.
Support correspondence is retained only for as long as reasonably necessary to respond, investigate an issue, protect legal rights, and satisfy applicable legal obligations.
Hugging Face and GitHub control the retention of their own technical request logs under their respective privacy policies. System translation language data and speech-recognition assets are managed separately by iOS or iPadOS and may remain in system storage according to operating-system behavior.
LocalMuse is not directed to children under 13. We do not knowingly collect personal information from children through an account, analytics system, cloud content service, or cloud speech-recognition service.
If you believe a child has provided personal information through a support request, contact us so the information can be reviewed and deleted where required.
Depending on where you live, you may have rights concerning personal information that we hold, including rights to request access, correction, deletion, restriction, objection, or a copy of that information.
Because LocalMuse creations, prompts, models, settings, and dictated prompt text remain on your device and are not available to us, you normally manage that information directly in the App or by deleting the App.
Requests concerning information provided to support can be sent to support@devserhii.com.
LocalMuse relies on the operating system’s app sandbox and Apple’s security controls to protect locally stored information. Voice input also relies on Apple’s system Speech framework and microphone permission controls.
No system can be guaranteed completely secure. Keep your device protected and its operating system up to date.
This Privacy Policy may be updated when LocalMuse, its models, prompt processing features, translation services, voice input features, speech services, or applicable legal requirements change. The updated version will be published on this page with a revised “Last updated” date.
For privacy requests, corrections, or deletion questions, contact support@devserhii.com.
You can also visit the support page.