Onedash uses artificial intelligence and machine learning to support selected features within the Services. Our current AI-powered functionality is primarily designed to run locally on the user's device rather than requiring Customer Data to be sent to an external AI provider. This page explains how those features work, what information they process, how local processing differs from future cloud or API-based AI features, and how Customer Data may be used in connection with AI development and improvement.

Our AI capabilities will continue to evolve. Where we introduce a feature that uses an external AI or model provider, we will update our applicable privacy, subprocessor, and transparency information to reflect that processing.

1. AI Features

Onedash currently uses local machine learning models for selected speech and productivity features.

• (a) speech-to-text and transcription features process audio to generate text and notes;
• (b) text-to-speech features process text to generate spoken audio;
• (c) speech-related functionality may be used to support voice-controlled workflows and related local application features; and
• (d) additional local AI or machine-learning functionality may be introduced as the Services develop.

These features are intended to assist users with productivity and accessibility. AI-generated or machine-generated output may contain errors and should be reviewed where accuracy is important.

2. Local AI

For AI features described as local, the relevant model runs on the user's device. The information required for that inference is processed by the local Onedash Application rather than being sent to an external AI model provider for the purpose of generating the result. For example, where local transcription is used, the speech-recognition model processes the applicable audio on the device to produce transcription output. Where local text-to-speech is used, the speech-synthesis model processes the applicable text on the device to produce audio.

Local AI processing does not mean that every item of data associated with the feature can never leave the device. If a user separately enables or uses a cloud, synchronisation, account, backup, or storage feature that transmits the resulting content, that separate processing is governed by our Privacy Policy, Data Processing Addendum, and the applicable Service settings. The use of local models reduces the amount of information that needs to be transmitted to external infrastructure for AI inference and allows certain features to operate without sending their inputs to a third-party AI provider.

3. AI Models

Certain local AI features require model files to be downloaded to the user's device. Those model files may be provided by Onedash or obtained from third-party model developers or repositories under applicable licence terms. Model files are used by the Application to perform local inference, and downloading a model does not by itself provide the model developer with access to Customer Data processed locally by Onedash.

Onedash may update, replace, add, or remove supported models as we improve performance, language support, compatibility, accuracy, security, or resource usage. Where a feature allows the user to select between models, languages, or voices, the selected model may be downloaded and stored locally for use by that feature.

4. Future Cloud AI

Onedash may introduce AI features that use external model APIs or cloud-based AI infrastructure. These features may include functionality such as drafting, summarisation, classification, search, extraction, assistance, or other generative and analytical capabilities. Where an AI feature uses an external provider, Onedash may need to transmit information relevant to the request to that provider so that the requested output can be generated. The information transmitted will depend on the feature and may include the user's prompt, selected content, relevant context, or other information required to perform the request.

Before or when such processing is introduced, Onedash will identify the relevant provider where required, update our subprocessor or service-provider information as appropriate, and describe material changes to the way Customer Data is processed. We do not currently represent that all future AI functionality will operate locally. Whether a feature is local or cloud-based will depend on the feature, technical requirements, user configuration, and the Services available at that time.

5. Customer Data and AI

Customer retains its rights in Customer Data as described in the Terms. The use of an AI-powered feature does not transfer ownership of Customer Data to Onedash. Where Customer Data is processed to provide an AI feature, that processing remains subject to the Terms, Privacy Policy, Data Processing Addendum, and any additional terms that apply to the relevant feature.

Onedash may create de-identified or aggregated information from Customer Data for Training Datasets as permitted by the Terms. Training Datasets are designed so that they do not identify Customer or include Personal Data where they are treated as de-identified data. Onedash may use Training Datasets for lawful business purposes, including data analysis, evaluation, testing, product development, and training or improving Onedash artificial intelligence and machine-learning systems. This Training Dataset process is separate from the ordinary operation of a local AI feature, and using local transcription or text-to-speech does not mean that the local model automatically retrains itself on the user's content.

6. AI Training Opt-Out

Customer may opt out of having Customer Data used to create future Training Datasets in accordance with the Terms. An opt-out applies prospectively within a reasonable period after the request is received and verified. An opt-out does not require Onedash to:

• (a) reverse model training that has already been completed;
• (b) remove information that has already been irreversibly de-identified or aggregated so that it no longer constitutes Personal Data; or
• (c) reconstruct or alter a model where the effect of previously de-identified training information cannot reasonably be isolated.

Customers with an applicable Business Associate Agreement are automatically excluded from Training Datasets unless otherwise expressly agreed in writing. Requests to opt out may be submitted to [email protected].

7. AI Inputs and Outputs

AI and machine-learning systems may produce incorrect, incomplete, misleading, or unexpected output. Output should not be assumed to be accurate solely because it was generated by an AI-powered feature. Customer is responsible for reviewing AI-generated output before relying on it where the output may affect legal, medical, financial, employment, safety, compliance, or other significant decisions. Onedash does not represent AI-generated output as professional legal, medical, financial, or other regulated advice.

Where applicable law permits, Onedash does not claim ownership of Customer-specific AI output generated for Customer. Similar or identical output may be generated for multiple users because AI output is not necessarily unique.

8. Safety and Misuse

AI features remain subject to the Onedash Acceptable Use Policy. Customer must not use AI functionality to facilitate unlawful, fraudulent, abusive, harmful, or otherwise prohibited activity. Onedash may apply technical controls, usage limits, filtering, validation, monitoring, or other safeguards to AI features where reasonably necessary to maintain security, service integrity, legal compliance, or appropriate use. Where future AI functionality is supplied through a third-party provider, additional provider-level safety or abuse controls may also apply.

9. Human Review and Automation

Current local speech and productivity features are primarily assistive. They generate transcription, audio, or other output for the user rather than independently making significant decisions about individuals. If Onedash introduces AI functionality that makes or materially contributes to automated decisions with significant effects on individuals, we will review the applicable transparency, privacy, and legal requirements before making that functionality available. Where appropriate, Onedash may identify AI-powered functionality within the user interface so users can understand when an output or interaction has been generated or assisted by AI.

10. Quick Reference

Does Onedash currently send local AI requests to an external AI provider?

No. For features described as local, the applicable model runs on the user's device and performs inference locally.

Does Onedash plan to use cloud or API-based AI?

Onedash may introduce cloud or API-based AI functionality in the future. Where that occurs, we will provide appropriate transparency regarding the provider and relevant data processing.

Does local AI automatically train on my content?

No. Local inference and model training are separate processes. Using a local AI feature does not mean that the model automatically retrains itself using the user's content.

Can Onedash use data to improve its AI?

Onedash may use de-identified or aggregated Training Datasets as permitted by the Terms. Customers may opt out of the creation of future Training Datasets in accordance with those Terms.

Does Onedash claim ownership of my Customer Data?

No. Customer retains its rights in Customer Data, subject to the limited licences and processing rights described in the Terms and other applicable agreements.

Are AI outputs always accurate?

No. AI-generated output may contain errors and should be reviewed before it is relied upon where accuracy is important.

11. Policy Updates

Our AI features, models, providers, and processing practices may change as Onedash develops. We may update this AI Transparency page to reflect new features, changes to local or cloud processing, new AI providers, changes to Training Dataset practices, or changes in applicable legal and regulatory requirements. Where a material change affects how Customer Personal Data is processed, we will provide notice where required by applicable law or our agreements.