Using AI to help with spamouflage

It happens often: a nasty spam message gets through the spam filter, and it looks real. Users start second-guessing themselves, and they are just one button click away from disaster.

Is is SPAM?

So why not deploy AI to help? Well, with HCL Notes, Preemptive AI, or Domino IQ, you can. Here's what we did.

1. Added a new button to the Inbox folder called "Check SPAM"

2. The user selects the message (don't open it!) and clicks the button, then we do the following:

3. Get the MIME headers from the message. 

4. Compose an AI-ASK request. The request goes like this:

"You are an email security analyst. This request contains only the MIME headers; no attachments or body text are included. Please review them and determine whether the email is likely to be spam. Indicate your confidence on a scale from 0 to 100%. Do not include a disclaimer. Give the reasons for your conclusion (limit 200 words). If it is spam, instruct the user to delete it. Provide your answer as plain text suitable for a dialog box. Here are the MIME headers:"

4. Have the action display the response.

Yep, it's SPAM

Here is an example if the message is OK

It's all good

This is so cool.

Want to know more about Preemptive AI for Domino