Automated Content Review & Moderation

Maus-Tec LLC, doing business as Maus-Tec Electronics


Our Commitment

Maus-Tec uses automated tools, including AI-based classifiers, to review content uploaded to our platform services. We believe you deserve to know exactly what these tools do, what they cannot do, and who operates them.

This document applies to all Maus-Tec platform services that accept user-uploaded content, including Maus-Hub and any future services.


What Our Tools Do

When you upload content to a Maus-Tec platform service, it may be processed by one or more of the following automated tools:

Audio Uploads

StepWhat HappensWhat It Produces
TranscriptionAudio is converted to textA text transcript of spoken words
Speaker CountingAudio is analyzed to count distinct speakersA number (e.g., “2 speakers detected”)
AI Voice DetectionAudio is checked for synthetic/AI-generated voicesA confidence score (real vs. AI-generated)
Content ClassificationThe text transcript is checked against our content policyA category label (approved / flagged for review)
Human ReviewFlagged items are reviewed by a Maus-Tec staff memberA final approve/reject decision

Image Uploads

StepWhat HappensWhat It Produces
Content ClassificationImage is checked against our content policyA category label (approved / flagged)
AI Image DetectionImage is checked for AI-generated contentA confidence score (real vs. AI-generated)
Human ReviewFlagged items are reviewed by a Maus-Tec staff memberA final approve/reject decision

Text Uploads

StepWhat HappensWhat It Produces
Content ClassificationText is checked against our content policyA category label (approved / flagged)
Human ReviewFlagged items are reviewed by a Maus-Tec staff memberA final approve/reject decision

What Does NOT Happen

We want to be absolutely clear about what our tools cannot do:

  1. No uploaded content is used to train any AI model. All third-party services we use contractually guarantee that data submitted through their APIs is not used for model training. This is a legally binding commercial agreement, not a policy statement.

  2. No voice cloning is possible from our pipeline. Our transcription tool converts speech to text — like a fast typist listening to a recording. It cannot generate speech, synthesize voices, or produce any audio output.

  3. No voiceprints are stored. Speaker counting produces temporary numerical statistics (like a word count for voices) that are discarded after analysis. These numbers cannot reconstruct a voice.

  4. No AI listens to your audio. Our content classifier receives only the text transcript. It never receives, processes, or “hears” any audio file.

  5. Maus-Tec does not operate any generative voice AI. We do not build, train, host, or sell voice synthesis, voice cloning, or text-to-speech technology.

  6. No content is shared between users. Each upload is processed in isolation. One user’s content is never used to inform, compare against, or improve analysis of another user’s content.

  7. No automated decision is final. Content flagged by automated tools is always reviewed by a human Maus-Tec staff member before any action is taken against your account.


Who Processes Your Content

We name every third-party service that may process your content:

WhatService ProviderTheir RoleYour Data Sent
Audio transcriptionSelf-hosted (open-source model) or third-party APISpeech → text conversionAudio file
Speaker countingSelf-hosted (pyannote-audio, open source)Count distinct voicesAudio file (never leaves our servers)
Content classificationAnthropic (Claude)Categorize text and images against content policyText transcript or image only — never audio
AI voice detectionTo be determinedDetect synthetic/AI-generated voicesAudio file

We will update this table whenever our vendor relationships change.

Self-Hosting Commitment

We are actively working to run all content processing on our own infrastructure. Our goal is that no uploaded audio ever leaves Maus-Tec servers. Speaker counting already runs entirely on our infrastructure using open-source software. We are evaluating self-hosted options for transcription and content classification as well.


Human-Only Review

If you do not want automated tools processing your content, you may opt into human-only review from your account settings.

When human-only review is enabled:

Honest Tradeoffs

Automated + Human ReviewHuman-Only Review
Turnaround timeMinutes to hours3–5 business days
ConsistencyHigh (same criteria every time)Variable (human judgment)
AI voice detectionAutomated screeningHuman ear only
PrivacyContent processed by tools listed aboveContent reviewed by a human employee
Cost to youNone (included)None (included)

Human-only review is a genuine option, not a second-class path. There is no additional fee or penalty for choosing it.


Data Handling

Transcript Retention

Text transcripts generated from audio are deleted within 24 hours of moderation completion. Transcripts are not stored, indexed, or searchable after that period.

Processing Isolation

When using third-party APIs, we send the minimum data necessary:

Audit Trail

We log which tools processed each upload and when, so we can respond to your inquiries about exactly what happened to your content.


Your Rights

Right to Know

You may contact us at any time to ask what automated processing was performed on a specific upload. We will provide:

Right to Human Review

You may request human review of any automated decision at any time. To appeal an automated decision, use the appeal form in your account dashboard or contact us directly.

Right to Opt Out

You may enable human-only review at any time from your account settings. This applies to all future uploads. Previously processed content is not reprocessed.


Our use of automated content review is based on:


Frequently Asked Questions

“Does my audio train any AI?”

No. Absolutely not. Our third-party API providers contractually guarantee that customer data is not used for model training. We can provide the specific contractual language upon request.

“Can these tools clone my voice?”

No. Our transcription tool converts speech to text. It has no voice synthesis capability. Our speaker counting tool produces numerical statistics, not voice recordings. None of our tools can generate, synthesize, or clone any voice.

“Why not just have humans review everything?”

A small team cannot manually review every upload in a timely manner. Automated tools let us give you faster approval times, apply consistent standards, and keep platform fees low. We use automation as a first-pass filter — anything flagged is always reviewed by a human.

“What if I disagree with a moderation decision?”

You may appeal any decision. Use the appeal form in your account dashboard or email us at info@maus-tec.com. A different human reviewer will evaluate your appeal.


Changes to This Policy

We will update this document when our automated processing tools or vendor relationships change. Significant changes will be communicated to users through the platform.


Contact

For questions about our automated content review practices:

Email: info@maus-tec.com