SecurityGateway includes AI email classification for stronger protection against email-borne threats.

AI email classification is a threat detection feature that uses artificial intelligence to analyze message content and classify messages as spam, phishing, commercial, containing personally identifiable information, or legitimate. You can then set rules to determine which messages will be analyzed and what actions will be taken (such as reject, quarantine, or add to the message’s spam score) based on the resulting AI classification.

The benefits of using AI email classification include:

• Advanced Threat Detection to Identify sophisticated phishing attempts that traditional filters might miss.
• Sensitive Content Identification to detect messages containing Personally Identifiable Information (PII) or other sensitive data.
• And Improved filtering with greater accuracy.

Watch our video to learn more about how to configure AI email classification in SecurityGateway.

How AI Email Classification Works

To understand how AI email classification works, it’s helpful to understand these three concepts: AI models, prompts, and rules.

AI Models

An AI model is a process that uses artificial intelligence techniques such as machine learning or deep learning, to perform tasks that typically require human intelligence. These tasks include understanding natural language, recognizing images, and making predictions. AI models are trained on large datasets and can learn patterns, relationships, and decision-making processes to perform these tasks autonomously.

Prompts

An AI prompt is a specific instruction, question, or piece of text that is passed to your selected AI model to guide its behavior and generate a desired output. It's essentially a way to communicate with the AI service and tell it what you want it to do.

Rules

Rules allow you to specify which email messages you’d like to analyze, and what actions you’d like to take, such as quarantine or reject, based on analysis results.

How to Use AI Email Classification in SecurityGateway

The steps to implement AI email classification in SecurityGateway are:

1. Add an AI model to query
2. Configure AI prompts
3. Define rules to determine which messages to analyze & what actions to take based on analysis results

Step 1: Add an AI model to query

The first step is to define an AI model to query. You have four AI models to choose from:

OpenAI/ChatGPT
- Gpt-4.1-mini may provide a good balance of performance and cost for general tasks
- GPT4.1 is more powerful and, while it may be better for complex analysis, it may incur higher costs 
Google Gemini
- Designed for speed and accuracy
Local
-  Use if you prefer not to send data to a third-party provider
Custom

Step 2: Configure Prompts

The next step in configuring AI classification in SecurityGateway is to configure prompts. An AI prompt is a specific instruction, question, or block of text that is passed to your selected AI model to guide its behavior and generate a desired output. As I mentioned above, it allows you to communicate with the AI service and tell it what you want it to do.

This prompt will request the AI model to return one of the allowed classification labels specified - for example LEGITIMATE, UNSOLICITED, COMMERCIAL, PHISHING, OR HIPAA, among others.

Step 3: Define Rules

The next step is to define AI Rules to specify which email messages you’d like to analyze, and what actions you’d like to take, such as quarantine or reject, based on the results of the email classification process. For example, you could choose to use AI Classification only for messages addressed to certain users, such as those whom you think may be most vulnerable to phishing attempts.

Cost & Testing Considerations

Because data is passed to a third-party provider, cost is a valid concern when using cloud-based AI models, therefore you should test various models to see which would work best for you.

• OpenAI may offer a certain number of free tokens for new accounts. For more extensive testing or if you agree to their data usage policies (which might involve allowing them to use your data for training purposes), you might find options for reduced-cost or free testing.
• Google Cloud AI, when used with a free Google Cloud account, often provides a limited number of free requests as part of their free tier, which can be useful for initial testing and familiarization.

To avoid any unexpected costs, always review the current terms of service for any AI provider.

AI email classification is one of many security features found in SecurityGateway. Whether you’re using Microsoft 365, Microsoft Exchange, MDaemon, or any other email service, SecurityGateway can help protect your business from email-borne threats. Click here to download your free trial!