Some CoreView tools, such as the Script Optimizer, leverage generative artificial intelligence. Here we describe the security and privacy measures implemented:
How is your AI technology built?
Let’s talk about the tech behind it—our AI infrastructure is built with enterprise use in mind.
- AI Provider: our AI is powered by Azure OpenAI.
- Hosting: the model is hosted within Azure Datacenters located in the European Union Economic (EUE) region.
- Model used: we use GPT-4o, a state-of-the-art Large Language Model (LLM) tailored for enterprise applications.
- User Interface: our customized CoreChat UI provides a familiar and intuitive experience similar to ChatGPT, but it’s tailored specifically for CoreView functionalities.
How is your data handled with privacy in mind?
Concerned about data handling and privacy? Rest assured, we’re committed to ensuring the privacy and security of your data.
- Our delivery model involves a private implementation within CoreView’s infrastructure to keep processing secure.
-
Invocation process: when it comes to interacting with AI:
- The base models (Azure OpenAI) are directly invoked via the user interface.
- Our custom models use an internal API that’s crafted using an open-source agentic framework.
- Infrastructure: hosted on Microsoft Azure but kept logically separate from CoreView’s production environment and your customer data.
How do you uphold security and compliance standards?
Security and compliance are top priorities for us. We’ve got strict measures in place to ensure content and data are handled carefully.
Content and data segregation
- Content type: the Custom Action Optimizer works exclusively with text-based PowerShell inputs and outputs.
- Data retention: customer data is not used for training models. Management aligns with Microsoft OpenAI Data Privacy & Security policies.
-
Segregation controls: for content and data segregation:
- Sensitive data handling: we’ve implemented logic-based filters to prevent processing of sensitive data.
- Prompt management: no prompts or conversations are stored, thanks to Azure AI’s segregation solutions..
- Access controls: access to AI responses is restricted within your specific organizational context and session.
How do you ensure ethical AI practices and maintain quality control?
AI ethics and quality standards are integral to our processes:
- Built-in AI ethics controls: includes content moderation tools to prevent harmful or biased outputs.
- Bias and hallucination mitigation: we’re planning to introduce more checks and validations to mitigate bias or hallucinations in our AI outputs.
- Intellectual property rights: any outputs comply with the intellectual property framework outlined in Microsoft OpenAI service agreements.