Content Filtering More Information

Step 1
Troomi recommends you start using our AI’s Default setting associated with your child’s age. After experiencing how Troomi Content Filtering works for a few days, then make adjustments according to your child’s maturity level.

Step 2
Set Blur Flagged Content On or Off. When it is On, all sent and received “Flagged” content will be blurred by Troomi’s AI.

What does your child see?
Flagged words will be replaced by (***).

Flagged images will be blurred.

Emojis will be blurred.

Videos and GIFs are treated like a series of images that are sampled once every few seconds. When they are flagged the phone will display a blurred image of the first frame of the video.

What do you see in the Parent Portal?
Content is not blurred in the Parent Portal. Parents will see all blurred content, so they can have meaningful conversations with their children, and help train Troomi’s AI.

Step 3
Set how often you would like to be notified, when your child sends or receives Flagged content.

We recommend you set this to “Daily,” until you get a feel for how many Notifications receive. The frequency options are As Soon As Possible, Daily, Weekly, Monthly and never.

Step 4
There are two ways to manage your Customize Flag/Allow Word and Emoji List and help train Troomi’s AI. We recommend you wait a few days after turning on before you do this.

Method 1
Troomi’s AI has thousands of words and emojis that get flagged because they are not appropriate for any child. There are others that may or may not be appropriate based on their age and maturity. If you discover that the AI missed a word or an emoji in a text that should have been flagged for the age/maturity of your child, you can add it to your child’s personalized Blocked and Allowed List.

From your Parent Portal Text Usage dashboard, you click on “View,” which will take you to a screen that emulates your child’s phone screen. You can then click on the flagged texts or images, and adjust how the AI rates them for you child.

Method 2
Another way to manage your text/emoji Flag/Allow List outside the context of a text, is by clicking on the “Customize Word and Emoji List” on the Content Filtering Settings page.

*Note: Regardless of what the Troomi AI recommends, words that are specifically Blocked using this list will ALWAYS be flagged and blurred and words that are specifically Allowed using this list will ALWAYS be shown to the child.

Step 5
Training the AI for images, videos, and GIFS sent/received is simple as well. Clicking on an image, video and GIF with your child’s text will pop up a screen that lets you coach the AI in two easy steps.

Helpful Information
Among the characteristics that our children, tweens, and teens are born with are curiosity, innovation, creativity, innocence, inexperience, and naivety. When a child grows up in a safe and nurturing environment, none of these characteristics are bad. However, the world is full of people who exploit these adolescent characteristics and cause harm. It is estimated that there are over 500,000 predators in the United States actively trying to harm our precious kids. Just curiosity itself among a group of teenagers can get them into trouble.

The good news is that Troomi empowers parents to choose who their kids receive and send texts and phone calls from on their Contact List. Generally, this works really well when a parent or guardian knows the individuals their kids interact with. Still, as they grow into tweens and teens, parents may not be as familiar with every person their child associates with.

In this world of texting and social media, there is literally no way for a parent to moderate their kids’ text messages, let alone understand them. Troomi’s AI was developed as a tool to assist parents in identifying potentially harmful messages sent/received by our Troomi kids so that parents can be notified and have the time and opportunity to intervene in situations before they cause their children harm.

Our Troomi AI smart filters will work even when bad actors actively try to circumvent them. Using AI and Natural Language Analysis and Proprietary Algorithms, our filters can detect millions of variations for each unwanted word: letter replacements, character insertions, phonetic variations, leet speak, obfuscation, hyphenation, special characters, embeddings secret emojis and emoji combinations, etc.

Before you continue, please know that some of the examples we share may be uncomfortable for parents to read. Here are a few examples of how bad actors or creative teenagers use language that parents and guardians may not recognize or that other filters miss. See how you do on this test created by Joe Yeager on pediatricsafety.net:

Below are examples of code words that, when used in context, may or may not have meanings that we would not want our kids to share. How many of these would you have caught?

BOB: An acronym for Battery Operated Boyfriend.

Example: She’s been spending a lot of time with BOB lately.

Bunny: An abbreviated form of a “Rope Bunny” – someone who likes being tied up.

Example: I’ve been hoping to find a cute bunny lately, but no such luck!

Chocolate: A black person.

Example: I’ve been craving chocolate a lot lately!

Headache: When a person, usually a male, is aroused and looking for sex.

Example: Man, I wish I could do something about this headache. It just won’t go away.

Little: A person who pretends to be much younger than they are chronologically. The difference can be years or even decades. This person is often in search of someone who is looking to take care of them (not always sexually).

Example: I woke up feeling very little today.

Mary Jane: Another word for marijuana. Also known as MJ.

Example: Has anyone seen MJ lately? I’m looking for her.

Pet: A person who likes to be cared for, often in a submissive role.

Example: I’m looking for the perfect pet. Anyone? [Done in a chat room]

Pizza: A euphemism for sex. The idea is that there is no such thing as bad pizza, and there is no such thing as bad sex.

Example: I really need to get some pizza today!

Smash: To have casual sex.

Example: Whenever I see him, I just want to smash him.

Troomi’s AI is here to help! It not only detects specific emoji combinations, words, and phrases, but it also detects millions of letter replacements, character insertions, phonetic variations, leet speak, obfuscation, hyphenation, special characters, embeddings, etc.

Content Filtering is not perfect, but it is cutting-edge, and with your help, it will improve. It is significantly better than not knowing what your tweens and teens are texting and sharing. On occasion, our AI may flag something incorrectly; these are known as false positives and false negatives. A false positive happens when the AI thinks something is inappropriate when it is not. A false negative happens if the AI misses something that you feel is inappropriate and doesn’t flag it. If this happens, you can coach the AI by checking a box that confirms the content is ok, or with one click, let the AI know the content was inappropriate for your child’s age and maturity.

As an organization, we feel strongly that Adult Content such as nudity, extreme violence, etc. and content that would suggest a child is being harmed or may cause harm to themselves will be flagged regardless of a parent’s settings.

Troomi has combined thousands of filter variations into four categories with filter settings that can be customized for each of your children. The aggregated filters are Profanity, Suggestive, Violence/Gore, and Drugs/Alcohol. Below is an example of how you can adjust our Profanity Filter for a tween.