At 360Learning, our commitment to privacy and security extends directly to the development of our artificial intelligence functionalities. We believe AI must be developed responsibly, with a strong focus on transparency, security, and ethics. To learn more, you can read our commitments to responsible AI here ↗
Our AI Technology & Content
What AI technology powers 360Learning's features?
At 360Learning, we use Microsoft Azure OpenAI Service to provide advanced AI capabilities. This service gives us access to powerful models like GPT-4, GPT-4o, and their variations, all with the robust security and enterprise-grade privacy features of Azure. We update these models frequently to ensure the highest quality output for you. Additionally, for our "Skills" service customers, we use custom Transformers models. These models are built on each customer's data and are combined with Azure OpenAI to deliver a truly personalized experience.
Where do AI-generated content and results come from?
AI results and content are generated using proprietary workflows designed by 360Learning. These workflows utilize several models from the Microsoft Azure OpenAI Service and are constantly updated to ensure high-quality output. The content is created directly from your inputs (prompts, uploaded documents) within our secure environment.
How is the data I share with AI used?
The data you share with our AI system is used only to produce the direct outputs you request. This includes:
- Content creation: Generating learning content, courses, text, questions, and quizzes.
- Administrative tasks: Facilitating translation, tagging, and platform support.
- Learner experience: Providing content recommendations, smart search, and adaptive learning paths.
- New learning methods: Providing feedback on learner answers.
Your inputs and the AI-generated outputs are handled in accordance with Microsoft Azure’s strict privacy and security provisions.
Data Privacy, Storage, and Ownership
Does the AI system store the data I share? Where is it stored?
Yes, the data you share, such as prompts, is temporarily stored for a maximum of 30 days. This is done for the purpose of providing the services of the contract. This data is stored within secure Microsoft Azure Data Centers in Europe. We also have the capacity to store data locally based on customer needs.
Is customer data used to train the OpenAI models?
No. We do not fine-tune or train the underlying OpenAI models using your data. The models we use are off-the-shelf models from Azure OpenAI, accessed via their API. This means that your data is:
- Not available to other users.
- Not available to OpenAI.
- Not used to improve OpenAI’s models.
- Not used to improve Microsoft or any third-party products or services.
The data flows are only inbound to provide the requested service. No data from the AI's public knowledge is shared back. Your data is exclusively used for the performance of the services provided by 360Learning.
Will my data be shared with third parties?
No. Your data is exclusively used for the performance of the services provided by 360Learning. We do not share your data with any third parties. Microsoft Azure is a trusted sub-processor that processes your data strictly for the purpose of providing the Azure OpenAI Service to us, under our existing contract and Data Protection Agreement.
How do you manage the segregation and privacy of customer data?
Your data is kept strictly segregated. Each customer platform has its own dedicated systems, and there is no data mixing across platforms. Data is processed and stored on Microsoft Azure data centers. Access to AI functionalities and the data used is controlled via role matrices, ensuring that only authorized users can access sensitive information. Customer data remains owned by the customer, is not sold or disclosed to third parties, and is pseudonymized in line with GDPR requirements.
Does the customer own the content generated from their inputs?
Yes, you, the customer, own all the data and content generated from your inputs within the 360Learning platform. This includes any content created using our AI features. This content is created for your exclusive use within your platform.
AI Functionality & Quality Assurance
How do these AI functionalities fit into the platform?
Our AI functionalities are seamlessly integrated to enrich your content creation and learning experience. At this time, our generative AI features are used for:
- Content Creation: Generating courses, text, and quizzes from your documents.
- Administrative Tasks: Assisting with translation and tagging.
- Learner Experience: Providing personalized content recommendations and smart search capabilities.
- New Learning Methods: Providing feedback on learner answers.
How is the data processed to build content recommendations?
The model used for generating content recommendations is LightFM, an open-source model. To comply with data privacy regulations, the model has been trained with anonymized data, including: User ID, Item ID (for courses or learning paths), Item Language, Group ID, and Tags ID. Each customer platform has its own dedicated model, trained on its own proprietary data. This ensures there is no data mixing across different platforms.
To ensure quality, the model is not self-training. Every update we make is a standard evolution with close monitoring and testing before release
What types of content can your AI process?
Currently, our AI functionalities are able to handle text-to-text generation. Our team is actively exploring multi-modality to support inputs and outputs such as speech, images, and videos in a near future.
How do you ensure AI-generated content is correct and error-free? How do you avoid bias and hallucinations?
We have implemented several measures to ensure the quality and safety of AI-generated content:
- Content Filters: We use the "high" filters and control functionalities provided by Microsoft Azure OpenAI.
- Monitoring: We closely monitor outputs and block non-compliant content. Our systems include workflows to detect sensitive content.
- Review: While we strive for accuracy, errors and mistakes can happen. Users can see a disclaimer in courses that include AI-generated content. We strongly encourage users to review and verify all generated content before publishing.
- Compliance Verification: We have implemented a compliance verification feature accessible to platform administrators for additional oversight.
- Bias Mitigation: We use Azure OpenAI's native content filters to reduce bias and have workflows to ensure alignment with our ethical and legal standards.
How do you monitor and track performance, reliability, and bias?
We have developed an advanced monitoring dashboard with our partners Datadog and Microsoft. This system sends real-time alerts on issues related to bias, performance, and reliability, allowing us to proactively address any concerns and ensure a consistent user experience.
Legal & Governance
Is 360Learning’s use of AI GDPR-compliant?
Yes. We use Microsoft Azure OpenAI technology, and Microsoft Azure is already one of our GDPR-compliant sub-processors. The use of its OpenAI services does not alter the contract and Data Protection Agreement (DPA) provisions we have in place.
Are you aware of the EU AI Act? Have you conducted an AI conformity assessment?
Yes, we are closely following the latest regulations, including the EU AI Act. With the support of our in-house legal experts and external counsel, we have conducted an analysis and determined that our AI system falls into the category of Limited Risk AI System, as it does not pose significant risks to individuals' fundamental rights. We are actively following regulatory guidelines to ensure full compliance within established deadlines.
Have you put in place internal governance related to the use of AI?
Yes. We have implemented an internal charter on AI usage that establishes clear guidelines for the responsible and ethical use of artificial intelligence within our group. We are also developing supplementary training programs and documentation to educate our employees on best practices, as part of the AI Act literacy obligations.