United States
AI You Can Trust
Built for accessibility and putting you in control
This statement is for customers, partners and Dolphin tech users to learn how Dolphin uses AI responsibly in its products and services
About AI in Dolphin products
Artificial intelligence (AI) is an important part of how customers experience Dolphin products. AI can help you work more efficiently and access features designed to support your independence in education, the workplace and at home, all with accessibility at the core.
Dolphin uses OpenAI GPT models and Google Gemini to process text and images. We also use OpenAI text embedding models to support semantic search across manuals, knowledge base articles, and training materials, helping users find relevant information efficiently and accurately.
When you are using Dolphin products, you can expect AI to be used responsibly, with transparency, control and accountability.
Five key principles for AI use
Dolphin products use AI to support and empower you, without compromising your trust. We focus on features that are safe, transparent and truly helpful.
Dolphin is guided by these principles:
- Empowering accessibility: Tools are designed to help you learn more effectively and become more independent.
- Focussed on your content: AI works only with the content you choose to provide, rather than answering general questions.
- Giving you control: You decide when and how AI features are used, with clear visibility throughout.
- Designed for accuracy: Features are tested for accuracy and improved through ongoing feedback and monitoring.
- Protecting your data: Your content is not used to train AI models. We only store your data if you choose to submit feedback.
Empowering accessibility
AI is integrated into our products to help our customers to work more independently, learn more effectively and confidently achieve more.
Our AI-enabled features enable you to:
- Find answers effortlessly, asking product‑related questions in natural, everyday language.
- Understand visual information such as charts, diagrams, photos and handwriting, using natural language to explore and query images.
- Boost comprehension, with tools that summarise complex text, reinforce learning through quizzes and flashcards. Our approach empowers enquiry-based learning - encouraging students to think critically and discover answers for themselves.
Focussed on your content
We believe an integrated approach to AI is more responsible than relying on mainstream AI tools. By designing AI that works within our products, we focus on removing accessibility barriers and helping our customers be more productive. We ensure the AI tools required are quick, easy and safe to access.
Our design ensures AI services only receive the specific content it is provided by you, such as an image, a textbook chapter, or a help request. This ensures the technology is both purposeful and protected against misuse, to give users the benefits of AI without the risks associated with open systems.
Giving you control
We clearly communicate when AI is being used and what it does.
Users will always be informed when an AI system is assisting them. This enables you to make informed decisions whenever you use an AI-enabled feature.
Designed for accuracy
We evaluate AI features for accuracy, reliability and safety. Our reasonable precautions include:
- Robust internal testing
- Monitoring user feedback to assess quality and identify unexpected or harmful outputs
- Continuously improving systems based on real-world usage and emerging risk
While we take all reasonable precautions, no AI system can be guaranteed to be fully accurate, safe or reliable in all circumstances. Large language models (LLMs) can produce incorrect or unpredictable outputs, and some risks cannot be fully mitigated at the application layer. We rely on the underlying LLM providers for core safeguards and we apply additional controls where possible.
Users should always apply their own judgement when interpreting AI generated content.
Protecting your data
We treat user data with the highest level of care. AI features are designed so that data is processed securely and only for the specific purpose of delivering the requested functionality.
Our licenses are with trusted AI providers. This ensures:
- Data isolation: Customer data is not used to train or improve public models. We do not store any data from the use of AI, except where users choose to provide feedback to an AI response
- Security and compliance: Models operate within environments that meet strict enterprise security standards
- Reliability: We rely on providers with robust uptime, monitoring and operational safeguards
- Governance: Contractual commitments around data handling, privacy and responsible use
Investing in the AI future
At Dolphin, we are excited about the future of AI and the potential benefits it can bring to our customers in education, the workplace and at home. We are actively researching and designing future AI-powered solutions through our product development teams.
We recognise that Responsible AI is not a one-time effort. We continuously review user feedback and regularly review new AI capabilities and risks. Our goal is to ensure Dolphin AI tools are trustworthy, dependable and engaging, so they remain a beneficial part of our products - today and in the future.