Security, Privacy and Transparency

Overview

Volt is built with security principles and the EU AI Act in mind. This article provides detailed information about how Volt handles your data, the security measures in place, and our commitment to transparency.

How Volt Works

Volt is powered by large language models developed by OpenAI and Anthropic. These include OpenAI's GPT series and Anthropic's Claude series of models.

Volt uses these models to analyse and generate responses to prompts in natural language, providing relevant and contextual answers. These models generate responses based on your inputs and are probabilistic in nature—they predict the most probable next word or text based on their training data.

Architecture Flexibility

We've built flexibility into our architecture using well-supported frameworks. This allows us to:

• Test and switch between different LLMs as technology evolves

• Respond to different task types effectively

• Continuously improve based on real usage and feedback

• Adapt to new models without being locked into any single provider

Data Privacy and Security

What We Process

When you use Volt, we process:

• Your prompts (inputs) and generated responses (outputs)

• Relevant context from your instance, such as content pieces, surveys, one-to-ones, or documents from connected third-party products

• Interaction data about feature usage and collaboration patterns

• Optional feedback you provide about the feature

Data Protection Guarantees

Your prompts and responses:

Agreements with AI Providers

We have agreements with OpenAI and Anthropic that explicitly prohibit them from training models with customer data. Additionally, we maintain Zero Data Retention agreements with each provider, ensuring that any data sent for processing is immediately destroyed after completion.

This provides customers with strong assurance that these providers cannot access, analyse, or retain their data for any purpose beyond immediate processing.

Permission Awareness

Volt respects all existing platform permissions. You'll only receive suggestions for content you have access to. To control content visibility in responses to other users, work with your organisation admin to configure appropriate permissions.

Pre-requisites

To access Volt's security and privacy features, your organisation must have opted into the Volt beta programme.

Role Requirements

⚠️ Note: During the beta period, Volt is only available to Owners and Admins. Other roles do not have access.

Control and Customisation

Opt-In by Default

AI features are disabled by default. You have full control to enable specific Volt functionalities when your organisation is ready, allowing you to adopt AI capabilities at your own pace.

Features are granular—you can enable advanced search and content recommendations while keeping other features like manager coaching disabled.

Customisation Options (Roadmap)

Customisation is a key part of our roadmap. You'll be able to:

• Adjust the agent's tone and communication style

• Configure support levels (coaching through topics, using specific frameworks, follow-up quizzes, or direct answers)

• Tune the AI to reflect your organisation's tone, language, and approach

• Ensure AI responses align with how your organisation already communicates

Feedback and Oversight

User feedback is built directly into the experience. Users can:

• React to AI outputs inline

• Flag inaccurate or unhelpful responses

• View the reasoning behind AI responses, including source data where relevant

These signals help us continuously fine-tune agent performance. All AI interactions are logged for visibility and oversight.

Limitations and Best Practices

Known Limitations

Due to the probabilistic nature of language models, Volt may occasionally produce responses that are inaccurate, incomplete, or unreliable. Responses might not perfectly reflect source content or may include plausible-sounding but incorrect information.

Volt is less effective when:

• You need information not stored in or connected to Learn Amp

• You require completely consistent, invariable outputs

• Human oversight is unavailable

Best Practices

We recommend:

• Reviewing all responses before sharing them with others

• Being specific and detailed in your prompts

• Using human judgment to verify and refine outputs

• Understanding that Volt is a tool to augment, not replace, human decision-making

Compliance and Governance

We're committed to meeting EU AI Act 2025 requirements through:

• Transparency – Clear explanations, source references, and feedback tools

• Human Oversight – Capabilities for human review and intervention

• Granular Control – Fine-grained control over AI feature usage

• Audit Trails – Complete logs of AI interactions (stored for 365 days)

• Opt-In Controls – Clear consent mechanisms and usage records

• Data Governance – Comprehensive measures to protect customer data

All AI functionality requires explicit opt-in, and features are designed to meet classification requirements under the Act.

FAQs

Q: Does Learn Amp use my data to train its own products?
No. Volt does not use your inputs or outputs to train Learn Amp products and services. We only collect metadata about feature interactions.

Q: Is my data shared with other customers?
No. Your submitted data and received responses are used exclusively for your experience. We do not share data between customers.

Q: How long is interaction data stored?
Interaction logs are retained for 365 days for safety and security purposes. Storage durations align with our standard data retention policies and can be configured by the customer.

Q: Is data shared with LLM providers?
Some on-platform data is shared with our LLM providers solely to generate AI responses. However, strict agreements ensure this data is not retained or used to train their models.

Q: Can I disable Volt if we change our mind?
Yes. AI features can be disabled at any time through the Beta Features settings.

Troubleshooting

Last Reviewed: 27/11/2025