Overview
MoEngage’s Merlin AI suite is the future of productivity and content creation for marketing teams.
Our commitment is to provide an environment that ensures high availability and tight security for your content creation needs. We strive to maintain transparency about our security practices to ensure you understand our security-by-design approach. Merlin AI guides the generative strategy towards safety, security, and growth by utilizing the world's leading AI models and our internal proprietary algorithms.
Security isn't just a feature of our product, but an inherent part of our identity. It informs every decision we make and has been integral to our product since inception. This includes the data we collect, its usage, and the entities with whom we share it.
MoEngage Architecture
We strictly adhere to internal AI usage standards. Our design ensures customer data is kept separate from our model providers. Thus, no sensitive information is shared with 3rd party service providers at any stage. The intent of this policy is not to limit creativity or innovation but rather to ensure the usage of AI aligns with the company values and safeguards our customers' rights.
Merlin AI architecture allows for custom authentication protocols, complete control over access to global knowledge, and yet customization of generation, credits, and usage policy on an account level.
- We provide secure access through account login, which is accessible only to employees authorized for campaign creation, thereby preventing misuse. This means access and permissions are granted strictly based on employees' job requirements.
- As a built-in feature, Merlin AI offers advanced security and privacy measures, assuring you of data security and confidentiality without requiring additional setup or configurations. You can certainly be at peace, knowing that your information will always be secure with Merlin.
Merlin AI operates in collaboration with two systems that function asynchronously:
- Learning from client data: The intelligence within the first system is handled by MoEngage, which utilizes our proprietary algorithms to learn from past campaign data, their results, user actions, and customer preferences within a customer's workspace. This process strictly excludes the use of generative AI. For example, the Merlin AI copywriter leverages the Keyword Impact Quotient (KIQ) algorithm to glean insights into which keywords have historically been effective in campaign messaging within a workspace.
- Foundational Generative AI Models: For foundational models, we rely upon 3rd party tools. This system produces outputs for the users’ requests. Users can include suggestions informed by MoEngage’s algorithms drawn from previous campaigns and user data. Users may or may not choose to use the suggestions. Ensuring maximum confidentiality, we never share account details, business names, or any other personal information with our AI partners.
Our AI Partners
We take great pride in partnering with Microsoft Azure, Amazon Web service and Stability AI to provide industry-best privacy controls, giving you complete authority over your data.
The Amazon Web service creates a secure environment, particularly beneficial when handling sensitive information. The AWS emphasizes network isolation and facilitates managing and learning from the data point specific to your account e.g. keyword analysis. These learnings, then, without any specific PII or account information, can be passed on to Merlin AI by you as one of the input while using the feature.
Microsoft Azure deploys a stratified series of mitigations across four layers, following Microsoft's responsible AI standard. Technical provisions such as input and output content filtering aid in detecting harmful content. The latest research advancements have provided another layer of security such that the safety mechanism has been directly integrated into the generative models, improving the model's ability to manage harmful inputs and reducing the probability of generating a harmful response.
Stability.ai releases AI models as open code; allowing researchers and authorities to examine the model's operation for transparency. Likewise, the provided open datasets invite robust scrutiny for quality, fairness, and bias. When it comes to their applications and APIs, Stability.ai deploys comprehensive filters to process prompts and outputs, screening for any unsafe content that contradicts their platform's terms of service. Furthermore, Stability AI is introducing content authenticity norms and watermarking to help users and platforms identify content created with their hosted services.
| info |
Information Security features are inherently incorporated into Merlin AI models to avoid any misuse of AI. Depending on the specific use case, these safety protections are either built-in native to our partner models or independently implemented by us. Merlin AI does not exhibit any bias based on gender, race, age, or any other factors. Your prompts (inputs) and completions (outputs), your embeddings, and your training data:
Also, about open AI models, the Azure OpenAI Service is fully controlled by Microsoft. Microsoft hosts the OpenAI models in Microsoft’s Azure environment and the Service does NOT interact with any services operated by OpenAI (e.g. ChatGPT, or the OpenAI API) |
Your Responsibility
While in-build MoEngage implementation and our partners provide considerable security controls and uphold responsible AI standards, ensuring the appropriate use of the generated content ultimately resides with the customers.
Generative AI, while transformative, should not be relied upon to fully automate content creation, at least not yet. It must be leveraged as an assistive tool rather than a replacement for human judgment. Our policy and design underscores the need for human intervention and review, ensuring that content generated by AI is not published without evaluation for quality and accuracy.
Conclusion
Ensuring the security and privacy of customer information is vital to our company mission. The success of our customers is at the core of what we do.
- We and our customers hold the right to use the generated output for commercial purpose. However, it is suggested to manually review the generated content before using it in a campaign.
- As part of our ongoing enhancement practices, we may choose to either increase or decrease the date range of the data retained by Merlin AI. Please note, as explained above, neither of your PII data or campaign data-pointst are shared with our partners.
- We are steadfast in our commitment to safeguarding your data throughout the handling process. Your data belongs to you; it will never be sold or shared irresponsibly.
- To provide you the best possible content generation Merlin uses the best AI model possible for specific use cases. These 3rd party AI models are not trained on your data and will not be trained on your data.