AI Usage and Disclosure Policy
Last updated
Contents
1. Purpose
1.1This Policy explains how artificial intelligence is used in the SphereOne Services provided by Altrivis Management LLC ("Altrivis", "we" or "us"), what happens to data when AI features are used, the limits of AI output, and the rules that apply to Customers and Users. It forms part of the SphereOne Terms of Service.
2. AI Features in SphereOne
2.1SphereOne presently includes the following AI features:
| Feature | What it does | Provider |
|---|---|---|
| SphereOne Assist | Summarises conversations and threads, drafts replies and messages, and answers questions about content in the Workspace that the User is permitted to see | Google (Gemini API, primary) and OpenAI (fallback) |
| Catch-ups, suggested replies and rewrite | Summarises what a User missed in a conversation, suggests short replies to a message, and rewrites a draft at the User's request | Google (Gemini API, primary) and OpenAI (fallback) |
| Daily brief | Summarises a User's own unread activity, for that User only | Google (Gemini API, primary) and OpenAI (fallback) |
| Translation | Translates messages into the language a User chooses, when that User turns translation on | Google (Gemini API, primary) and OpenAI (fallback) |
| Semantic search | Finds content by meaning rather than exact words, using an index derived from Workspace content | Google (Gemini API, primary) and OpenAI (fallback) |
| Speech to text | Turns a User's speech into text for dictation and live captions in browsers that cannot do this themselves, and, on paid plans, transcribes the sound of video clips and voice messages | OpenAI |
| Sphere (Enterprise plan) | Carries out a task a User asks for in words, such as sending a message or creating a task. Anything that reaches other people waits for the User to confirm it | Google (Gemini API, primary) and OpenAI (fallback) |
2.2Whenever a User interacts with SphereOne Assist, the interface identifies that the response is AI-generated. Content produced by AI features and inserted into a message by a User is the User's message, and the User is responsible for it.
2.3AI features do not include meeting recording, facial recognition, emotion recognition or any biometric processing by Altrivis. Transcription is limited to the speech to text described in Section 2.1. The optional "Only my voice" setting checks, on the User's own device, that "Hey SphereOne" was said by that User; its voiceprint never leaves the device (see Section 5.7 of our Privacy Policy).
3. How Data Is Handled
3.1Access limited to permissions. AI features act on behalf of the User and can use only the content that the User is already permitted to see in the Workspace.
3.2Sent to AI providers. When a User makes a request, the request and the Workspace content needed to answer it are sent to our AI provider through its application programming interface. The provider generates a response and returns it to us.
3.3No training. We do not use Customer Data, prompts or outputs to train or fine-tune any AI model. Our AI providers process this data under business API terms that do not permit them to use it to train their models.
3.4Provider retention. Our AI providers may retain requests and responses for up to thirty (30) days solely to monitor for abuse, after which they are deleted.
3.5Semantic index. Semantic search relies on an index derived from Workspace content. The index is stored in our infrastructure with the same protections as the original content, is available only within the Customer's Workspace, and is deleted when the underlying content or Workspace is deleted.
3.6Our data practices are further described in our Privacy Policy, Data Processing Addendum and Sub-processor List.
4. Administrator Controls
4.1AI features are enabled by default for paid plans and can be switched off by the Workspace Administrator in Settings > Workspace > AI features. When AI features are switched off, no Workspace content is sent to AI providers and the semantic index is deleted.
5. Limitations of AI Output
5.1AI output is generated by statistical models and may be inaccurate, incomplete, out of date, biased or inappropriate, even when it appears confident. It may omit important points from the content it summarises.
5.2Users must review AI output before relying on it or sharing it. AI output is not legal, financial, medical or other professional advice.
5.3AI output generated for one User may be similar to output generated for others. We do not warrant that output is unique or free from third party rights.
6. No Automated Decisions About Individuals
6.1SphereOne's AI features produce suggestions for Users to consider. They do not make decisions about individuals, and SphereOne does not use AI to score, rank or profile individuals.
6.2Customers shall not use AI output as the sole basis for any decision that produces legal or similarly significant effects on an individual, including decisions about recruitment, termination, promotion, compensation, discipline, credit, housing or insurance. Where a Customer uses AI output as one input to such a decision, the Customer is responsible for meaningful human review and for compliance with all laws on automated decision-making and AI in employment, including, where applicable, New York City Local Law 144, the Colorado Artificial Intelligence Act, the California Consumer Privacy Act regulations on automated decision-making technology, Article 22 of the GDPR, and the EU Artificial Intelligence Act.
7. Prohibited Uses
7.1Customers and Users shall not use AI features:
- (a)to generate content prohibited by our Acceptable Use Policy;
- (b)to impersonate a real person, or to create content intended to deceive others about its origin where disclosure is required by law;
- (c)to attempt to extract the models, their system instructions or training data, or to circumvent safety measures; or
- (d)in breach of the usage policies of our AI providers, which are incorporated by reference: Google Generative AI Prohibited Use Policy and OpenAI Usage Policies.
8. Transparency Under the EU AI Act
8.1Our AI features are not high-risk AI systems under the EU Artificial Intelligence Act. In accordance with Article 50 of that Act, we inform Users when they are interacting with an AI system, and AI-generated content is identified as such in the interface.
9. Our Commitments
9.1We will: (a) keep the description of AI features in this Policy accurate; (b) give at least thirty (30) days' notice before adding a new AI provider that will process Customer Data; (c) not introduce any AI feature that processes biometric data, records meetings, or makes decisions about individuals without updating this Policy and our other legal documents in advance, and obtaining any consent required; and (d) review this Policy at least annually.
10. Reporting Concerns
10.1If you believe an AI feature has produced harmful, offensive or unlawful output, please report it to support@sphereone.ai.