AI-Powered Insights: BrightCanary Enhances Text Monitoring for Parents

Monitoring your child’s text messages can help you ensure their safety and well-being. But when your child receives thousands of texts per week, the practicality of reading their messages becomes overwhelming. What if you miss something important, like references to bullying, drugs, or self-harm? Good news: We’re parenting in the digital age, and there are tools available that can make text monitoring easier and more effective. 

BrightCanary’s new AI-powered text message summary feature helps parents stay informed, without having to read every individual message. The app’s advanced technology not only summarizes text threads, but it also provides insights into your child’s emotional well-being and tips on how to have meaningful conversations. We sat down with BrightCanary CTO and co-founder Steve Dossick to talk about the new feature and how it sets BrightCanary apart from other child safety tools. 

All about BrightCanary’s new AI-powered text monitoring tool

Rebecca Paredes: Can you explain how BrightCanary’s new text message summary feature works?

Steve Dossick: Our goal is to help parents stay informed about their children’s lives and alert them when there’s a concern. To do that, we first anonymize each conversation thread and ask AI to do a couple of things: 

• Emotional content categorization: We classify the thread’s emotional content using the American Psychological Association’s guidelines. This helps parents understand the thread’s mood, whether it’s Angry, Friendly, Worried, or another vibe.
• Identifying concerning content: Does the content contain any red flags, such as NSFW images or references to drugs or alcohol? We track nine categories of concerning content, and we’ve found the latest AIs are incredibly accurate at identifying these issues.
• Summarize conversations: Our app provides a brief summary of the conversation, focusing especially on recent parts so parents can avoid reading every message unless it’s necessary. 
• Parental coaching prompts: The AI generates questions for parents to help prompt discussions about specific content. For instance, if the thread mentions a party with alcohol, the parents might want to ask how to talk to their child without seeming judgemental — while still making it clear that teens should not drink at parties.

We use the latest Large Language Model (LLM) AIs and supply them with extensive parenting guidelines to tailor answers for parental support. So, we created a parenting-focused AI, and then asked it to help parents.

RP: How accurate is this feature in measuring emotions and summarizing texts?

SD: No computer system is perfect, but we were truly astonished at how accurately the latest LLMs are able to summarize content and interpret human emotions. It seems strange to ask a machine to report on human emotions, but these AI systems are incredibly capable of understanding and summarizing text written by humans.

RP: How does the AI handle potentially sensitive or alarming content in text messages?

SD: The AI is trained with vast amounts of internet content. It is able to recognize patterns in text messages and identify sensitive or alarming content. We also tell the AI the child’s age range, which helps maintain accuracy. 

RP: How do parents feel about text message monitoring?

SD: Parents want to know that their kids are staying safe, especially in their first journeys online. We want to provide guardrails so they’re ready for all the positives and negatives the internet can throw at them as they grow older and more independent. At the same time, privacy remains important, particularly for older children and teens.

After launching our text message monitoring feature, we quickly decided that we wanted to use the power of AI to summarize text threads — so they don’t need to take the time to read every last message while maintaining a level of privacy. 

Of course, if something truly concerning is going on with a child, parents can still read the full thread. Additionally, our Ask the Canary AI offers guidance on addressing specific issues, like handling bullying at school.

RP: How do you see this feature evolving in the future? Are there any additional functionalities you plan to add?

SD: We’ve deployed this feature for text message monitoring, but we plan to use it across all of our child monitoring features (YouTube, Google, Instagram, TikTok).  

RP: What makes this feature different compared to other text monitoring tools available?

SD: Most existing text and image monitoring tools were designed for user-generated content sites like Yelp or TripAdvisor. These sites are concerned that people are sharing reviews with profanity or concerning references. However, children’s text message conversations are drastically different from what other monitoring tools have been trained for. 

Kids send one-word replies, abbreviations (mb, wyd), and misspellings. Traditional content moderation models struggle with this. LMs, on the other hand, are able to use the context to generate real insights into meaning and emotional content, making them better suited for child safety.

The bottom line

Ready to get started with text message monitoring? BrightCanary offers the most comprehensive child safety app for Apple devices. Start your free trial today, or learn more about how to monitor your child’s texts on iPhone.